co , o , and the history of the greenhouse effect: …
TRANSCRIPT
The Pennsylvania State University
The Graduate School
CO2, O2, AND THE HISTORY OF THE GREENHOUSE EFFECT:
SELECT PROBLEMS IN THE EVOLUTION OF THE
EARTH’S ATMOSPHERE AND CLIMATE
A Dissertation in Geosciences and Astrobiology
by
Rebecca C. Payne
© 2020 Rebecca C. Payne
Submitted in Partial Fulfillment of the Requirements
for the Degree of
Doctor of Philosophy
December 2020
ii
The dissertation of Rebecca C. Payne was reviewed and approved by the following:
James F. Kasting Evan Pugh Professor, Department of Geosciences Dissertation Advisor Chair of Committee
Christopher House Professor, Department of Geosciences
Julie Cosmidis Assistant Professor, Department of Geosciences Raymond Najjar Professor, Department of Meteorology and Atmospheric Sciences
Mark Patzkowsky Professor, Department of Geosciences Associate Head for Graduate Programs and Research (Geosciences)
iii
Abstract
To understand the history of life on Earth, it is essential to understand the evolution of the
global climate that has maintained a relatively stable and habitable environment for life over the
last 4.6 billion years. Earth’s climate itself is the result of a delicate balance between the
composition of the atmosphere and the processes on the surface—in the oceans, and on the
continents—that interact with it.
Prior to the rise of atmospheric O2 at ~2.45 Ga, Earth’s climate relied on CO2 as a major
atmospheric constituent in order to keep Earth at a habitable temperature in spite of lower solar
luminosity. While climate models have estimated that at least several percent of CO2 in the
atmosphere would have been needed to maintain a habitable climate, geochemical data from
paleosols have largely predicted CO2 one or two orders of magnitude lower, though there is
considerable disagreement between paleosol studies. We argue instead that oxidized iron
micrometeorites from ~2.7 Ga offer a new point of comparison that is not subject to the same
potential biases as paleosols. Abundant CO2 could have oxidized iron micrometeorites during
atmospheric entry, meaning that preserved micrometeorites may offer a means of estimating
atmospheric CO2 in an atmosphere largely devoid of other potential oxidants. We estimate that an
atmosphere of at least ~23% CO2 would be sufficient to oxidize the micrometeorites and keep
surface temperatures warm. Using a climate model, we demonstrate that the new CO2 constraint
from micrometeorites also supports the idea that the partial pressure of N2 (and overall surface
pressure) was lower on the early Earth, in order for high CO2 to not cause surface temperatures to
conflict with evidence of glaciation in the late Archean.
iv
The emergence of atmospheric O2 after ~2.45 Ga, and its role as a major constituent during
the last ~541 Ma in particular, resulted in a fundamental shift in the greenhouse effect and
atmospheric CO2. Using a coupled one-dimensional photochemical-climate model, we find that
atmospheric O2 equal to or higher than 21% (its present-day abundance) slightly enhances
warming by other greenhouse gases like CO2 and H2O, by broadening their absorption lines to
increase effectiveness. But O2 exerts a minor effect on surface temperature at most; our findings
support the classical interpretation that changes in solar luminosity, CO2, and global geography
are the primary controls on global climate in the Phanerozoic. We apply this analysis specifically
to the early Cretaceous, at ~100 Ma, and argue that higher atmospheric O2 would support both the
hot temperatures of the Cretaceous as well as the needs of the massive terrestrial life (namely,
dinosaurs) that mark that time in geologic history.
We also consider carbon cycling on long timescales, using a box model of the atmosphere-
ocean-seafloor to study changes in CO2 and in C sources and sinks over the last 100 Ma. We
demonstrate that both seafloor weathering and continental weathering are important to climate
change since the Cretaceous. Seafloor weathering is closely tied to seafloor temperatures, and so
was likely to have been a more important C sink during the Cretaceous when global temperatures
were higher than today. We agree with recent analyses that have suggested that continental
weathering—especially an increase in continental weatherability over the last ~100 Ma due to
tectonic activity and uplift, as well as sea level change—has driven much of the CO2 cycling.
v
Table of Contents List of Tables ................................................................................................................................................................ vi List of Figures ............................................................................................................................................................ vii Preface ....................................................................................................................................................................... viii Acknowledgements ...................................................................................................................................................... ix Chapter 1. Introduction ............................................................................................................................................... 1 Chapter 2. Iron micrometeorite oxidation as a proxy for atmospheric CO2 during the Neoarchean .................. 7
1. BACKGROUND ....................................................................................................................................................... 7 2. MICROMETEORITE OXIDATION .............................................................................................................................. 9
2.1. Entry physics ................................................................................................................................................. 9 2.2. Micrometeorite oxidation chemistry ........................................................................................................... 12
3. PHOTOCHEMICAL CALCULATIONS ....................................................................................................................... 17 3.1. Methods ....................................................................................................................................................... 17 3.2. Results ......................................................................................................................................................... 19
4. DISCUSSION ......................................................................................................................................................... 20 4.1. Constraints on pCO2 from Neoarchean climate ......................................................................................... 20 4.2. Constraints on pCO2 from paleosols ........................................................................................................... 25 4.3. Other analyses of the Tomkins et al. micrometeorites ................................................................................ 27
5. CONCLUSIONS ..................................................................................................................................................... 28 APPENDIX 1: FULL DERIVATIONS AND MODEL SPECIFICS ....................................................................................... 30
A1.1. Full derivation of CO2 mass balance, Eq. (4) ........................................................................................... 30 A1.2: Derivation of Eq. 10 ................................................................................................................................. 32
Chapter 3. The response of the greenhouse effect, and surface temperature, to changes in atmospheric O2 during the Phanerozoic .............................................................................................................................................. 51
1. BACKGROUND ..................................................................................................................................................... 51 2. CONSERVATION OF NITROGEN ............................................................................................................................. 53 3. RADIATIVE FORCING CALCULATIONS ................................................................................................................... 57 4. CONVERGED MODEL CALCULATIONS ................................................................................................................... 62 5. DISCUSSION ......................................................................................................................................................... 67
5.1. Additional issues with the Poulsen et al. calculations ................................................................................ 67 5.2. Comparison with other 3-D models ............................................................................................................ 71
Chapter 4. Long-term changes in atmospheric CO2 and the carbon cycle throughout the Phanerozoic .......... 74 1. BACKGROUND ..................................................................................................................................................... 74 2. METHODS ............................................................................................................................................................ 81
2.1. Model development ..................................................................................................................................... 81 2.2. Activation energy calculations .................................................................................................................... 85
3. RESULTS .............................................................................................................................................................. 88 4. DISCUSSION ......................................................................................................................................................... 94
4.1. Marine controls on C cycling ...................................................................................................................... 94 4.2. The role of continental processes in the Phanerozoic C cycle .................................................................... 96
5. CONCLUSIONS ..................................................................................................................................................... 99 Chapter 5. Concluding Remarks ............................................................................................................................. 101 Bibliography .............................................................................................................................................................. 103
vi
List of Tables 2.1 Oxidation pressure range upper and lower bounds……………………….……….
11
A1.1 Reaction rates……………………………………………………………………...
37
3.1 Effect of O2 on CH4 and N2O column mixing ratio……………………………….
56
3.2 O2 effects for one-step and converged calculations……………………..…………
60
3.3 O2 effect on planetary albedo……………………………………………………...
62
3.4 Effect of accurate ozone profile for converged calculations……………………… 65
vii
List of Figures 2.1 Vertical profiles of major Archean atmosphere constituents……………………...
20
2.2 1-D climate model estimates for Archean surface temperature…………………...
24
2.3 Comparison of climate model/paleosol Archean pCO2 estimates………………...
26
A1.1 Photochemical model atmosphere setup………..………………………..………..
35
A1.2 Comparison of vertical profiles of Archean atmosphere composition……..……..
36
3.1 Flux for one-step Cretaceous calculations………………………..……………….
59
3.2 Flux for converged Cretaceous calculations………………………..……………..
63
3.3 Vertical profile of ozone for Cretaceous calculations……………………………..
64
3.4 Vertical profile of temperature for Cretaceous calculations………………………
65
3.5 Comparison of O2 estimates from various models………………………………...
70
4.1 Proxy error analysis; model-proxy disagreement…………………………………
79
4.2 K-T&C 2-box and our 3-box model schematics…………………………………..
83
4.3 Model comparison for saturation state, temperature, and marine C fluxes……….
89
4.4 K-T&C model “best fit” case……………………………………………………...
92
4.5 Our model “best fit” case…………………………………………………………. 93
viii
Preface
Chapter 2 is published in the Proceedings of the National Academy of Sciences (PNAS) as
of January 2020 (https://doi.org/10.1073/pnas.1910698117), with coauthors Don Brownlee and
James Kasting. I ran the modeling cases described in the chapter and wrote the analysis, and thank
both for their input in refining the calculations and contributions to some of the text.
Chapter 3 is published in the Journal of Geophysical Research: Atmospheres (JGR:
Atmospheres) in August 2016 (https://doi.org/10.1002/2016JD025459), with coauthors Amber
Young (née Britt), Howard Chen, David Catling, and James Kasting. I conducted the modeling
and wrote the analysis; I thank all coauthors for their input on the text.
Chapter 4 is unpublished as of writing this dissertation, and is a collaboration between
myself and my advisor Jim Kasting. I modified the source code, conducted the modeling, and
wrote the analysis, and thank Jim for his guidance and contributions to some of the text.
ix
Acknowledgements
Thank you to my friends and family, who have kept me sane, been a constant source of fun and
support, and listened patiently to my lengthy excited and/or frustrated discussions about rocks and
air over the last several years.
I would also like to thank all of the Kasting research group, past and present—especially Sonny
Harman, Peng Liu, Benjamin Hayworth, Aoshuang Ji, Jingjun Xi, Evan Sneed—for their help,
camaraderie, and insight.
Finally, I’d like to thank my committee—Chris House, Julie Cosmidis, and Ray Najjar—as well
as the coauthors who have contributed to the various chapters of this dissertation and constitute
the royal “we” herein: Don Brownlee, David Catling, Amber Young (neé Britt), and Howard Chen,
and especially my advisor Jim Kasting. This work wouldn’t have been possible without you. I am
beyond grateful for everything I’ve learned these past few years, and can’t wait to keep learning.
x
“There is nothing more deceptive than an obvious fact.”
—The Adventures of Sherlock Holmes: The Boscombe Valley Mystery, Sir Arthur Conan Doyle
“The entire universe has been neatly divided into things to (a) mate with, (b) eat, (c) run away from, and (d) rocks.”
— Equal Rites, Terry Pratchett
“It is sometimes said that scientists are unromantic, that their passion to figure out robs the world of beauty and mystery. But is it not stirring to understand how the world actually works—that white light is made of colors, that color is the way we perceive the wavelengths of light, that transparent air reflects light, that in so doing it discriminates among the waves, and that the sky is blue for the same reason that the sunset is red? It does no harm to the romance of the sunset to know a little bit about it.”
—Pale Blue Dot: A Vision of the Human Future in Space, Carl Sagan
1
Chapter 1.
Introduction
The evolution of the Earth’s climate over the course of its 4.6-billion-year history
represents a dynamic balancing act between solar radiation, the changing chemical
composition of the atmosphere, and the adaptive strength of the Earth’s greenhouse effect.
It is widely accepted that the Sun was less bright earlier in its life, with a total luminosity
of only about 70% that of today at the start of Earth’s history (Kump et al., 2000). But
geologic evidence suggests that the Earth remained relatively warm for the vast majority
of its history: there is evidence for sustained liquid water on Earth’s surface at ~4.3 Ga
from detrital zircons (Mojzsis et al., 2001; Wilde et al., 2001), and indicators of life as early
as ~3.7 Ga in the Isua supracrustal belt (Nutman et al., 2016). The evidence for a warm
climate in spite of weaker solar luminosity—a challenge referred to as the Faint Young
Sun paradox (Sagan & Mullen, 1972; Kasting, 2010)—establishes the need for a strong
early greenhouse effect during the Archean (~4 – 2.5 Ga) to maintain surface temperatures
conducive to life.
Studying the evolution of the greenhouse effect is further complicated by the role
of O2 in Earth’s atmospheric history. O2 was negligible in the early Earth’s atmosphere,
resulting in a generally reducing atmosphere, before its concentration began to rise in the
late Archean around ~2.5 Ga (Holland, 1994/2006), in a change referred to as the Great
Oxygenation Event (GOE). Over time, this changed the atmosphere’s redox state from
weakly reducing to weakly oxidizing. In the Phanerozoic (~541 Ma to today), O2 became
a major atmospheric constituent and now constitutes approximately 21% of the atmosphere
2
by volume. This major shift in atmospheric composition and redox state means that the
makeup of greenhouse gases present in the atmosphere would have needed to adapt
accordingly, and that effectively studying the Earth’s greenhouse effect through time
requires studying it in tandem with the rise of atmospheric O2.
Several gases have been assessed for their potential as the dominant greenhouse gas
in the pre-oxygenated atmosphere prior to the rise of O2. Sagan and Mullen (1972) favored
NH3, but NH3 is photochemically unstable even in a reduced environment and is quickly
broken down when exposed to UV radiation (Kuhn & Atreya, 1979). It is therefore unlikely
that it would have been able to build up to the concentrations needed for it to have been a
major contributor to early warming. H2 can act as a greenhouse gas through collisions with
N2 (Wordsworth & Pierrehumbert, 2013), and production of H2 from early serpentinization
and volcanism may have contributed enough to boost the atmospheric greenhouse effect.
However, H2 could only have acted as a greenhouse gas for a geologically brief period, as
it would have likely been converted by methanogens to another greenhouse gas, CH4. CH4
provides some of modern greenhouse warming, but it may have been even more prevalent
in the pre-oxygenated atmosphere prior to ~2.5 Ga, when it was possible to build up to
greater concentrations without being quickly oxidized. CH4 has a lifetime of only about 10
years in today’s atmosphere, but can have a lifetime up to 1000 times longer in an anoxic
atmosphere (Haqq-Misra et al., 2008). This, and the fact that CH4 likely had a stable source
via early methanogenic life (Kharecha et al., 2005), means that it may have been relatively
abundant during the Archean. However, CH4 could still only have been a minor
atmospheric constituent compared to CO2; when the ratio CH4 to CO2 exceeds ~0.1, CH4
begins to polymerize and form haze as it is photolyzed, resulting in a net cooling effect as
3
it absorbs and reemits incoming solar radiation. Therefore, while CH4 may have
contributed to the early greenhouse effect, it is unlikely to have been the primary source of
warming.
CO2 plays a central role in modern climate change (both natural and
anthropogenic), but it likely also dominated the early Earth’s greenhouse effect. CO2
outgassed from volcanoes would have accumulated to much higher levels than today in a
volcanically active, pre-oxygen atmosphere (Kasting, 1993), and a CO2 concentration just
100 times higher than today’s would have been sufficient to keep the early Earth habitable
with regard to surface temperature. While other reduced greenhouse gases would have been
oxidized following the GOE (removing their contribution to the greenhouse effect), CO2
would not be as affected by this shift in atmospheric composition. In fact, the Huronian
Snowball Earth event at ~2.4 Ga—likely triggered by the oxidation and large-scale removal
of atmospheric CH4 and subsequent weakening of the greenhouse effect (Bekker et al.,
2005)—was probably ended by a buildup of CO2 from volcanic outgassing (Kirschvink et
al., 2000; Kopp et al., 2005). In the geologic record, the glacial deposits of the ~2.48
billion-year-old Huronian supergroup are overlain by siderites and dolomites that indicate
an oxygenated surface ocean and massive biological CO2 consumption, respectively, after
the retreat of the global ice cover. Ultimately, it is probable that CO2 has been the dominant
greenhouse gas throughout much of Earth history, and that it—along with changing O2—
dictated much of the major shifts in the atmospheric evolution and climate history on Earth.
The central three chapters of this dissertation focus on select problems in climate
history related to the role of CO2 and O2 through time. The first examines a new potential
means of estimating ancient pCO2, using one-dimensional modeling to recreate the pre-
4
oxygen Archean atmosphere around ~2.7 Ga. The second chapter utilizes a similar
modeling approach, using a coupled 1-D photochemical model and 1-D climate model, to
study the influence of O2 on the greenhouse effect of the much more recent Cretaceous
about ~100 Ma. The final chapter applies a different modeling technique, considering how
the interplay of CO2, O2, and the C cycle have shaped climate over the course of the
Phanerozoic.
As previously mentioned, what we know about Earth’s past climate is the result of
careful juxtaposition of climate modeling and geochemical analysis. That said, information
on Earth’s climate becomes increasingly limited with age, and the data that exist are often
progressively more difficult to decipher. Paleosols—ancient soils containing carbonate
minerals—have long been the most widely used and cited Archean proxies for pCO2
(Sheldon, 2008; Driese et al., 2011). But the estimates from paleosols are hard to constrain,
and there is considerable disagreement between individual paleosol studies (Kanzaki &
Murakami, 2015) as well as between paleosols and climate models (Kasting, 2015; Catling
& Kasting, 2017). Tomkins et al. (2016) analyzed oxidized iron micrometeorites from the
late Archean, theorizing that they indicated a high (~21%) abundance of upper atmospheric
O2 at 2.7 Ga, even though total atmospheric O2 is thought to have been negligible until at
least ~2.5 Ga. Our analysis (in Chapter 2) indicates that these micrometeorites are more
likely a proxy for Archean CO2 and, possibly, surface pressure. If so, this may provide a
new point of comparison for geochemical proxies and climate models. We investigate
pCO2 during the Neoarchean using compositional data from the 2.7 Ga micrometeorites
collected by Tomkins et al., along with novel mathematical interpretations. Assessing the
5
potential value of new atmospheric proxies is vital to understanding how Earth’s
atmosphere and climate have evolved over time.
In thinking about the greenhouse effect across time, it is also important to evaluate
competing effects that gases may exert on total atmospheric warming. While proxies
indicate abundances, models can help us resolve the interplay between gases and how this
affects energy balance. We therefore turn to the Phanerozoic in Chapter 3, long after the
rise of O2 as a major atmospheric constituent. First, we do an in-depth study of the effect
of O2 on surface temperatures, with a focus on the Cenomanian (~100 Ma). We compare
our results to those of Berner (2006) and Poulsen et al. (2015), who had conflicting views
on the relative importance of CO2 versus O2 on Phanerozoic climate. We compare the
influence of O2 to that of CO2 on surface temperature, as well as the evolution of the
biosphere at that time. We argue that Berner’s assertion that CO2 and weathering were the
primary controls on Phanerozoic climate is robust, with large-scale changes in atmospheric
O2 accentuating the strength of the greenhouse effect.
Finally, we combine our work utilizing both geochemical proxies and climate
models to try to understand long-term climate evolution during the Phanerozoic. We assess
existing models of the global C cycle and atmospheric CO2. In particular, we consider
Berner’s GEOCARB model, the COPSE model by Bergman et al., and the more recent C
cycle model by Krissansen-Totten & Catling. We then construct our own box model to
study CO2 and carbon burial in the Phanerozoic, extrapolating relationships between data
and systems over time to better understand long-term changes in climate and the
atmospheric greenhouse effect. We also use this model to revisit calculations by Berner,
Krissansen-Totten & Catling, and others, of the activation energy for continental
6
weathering and use it to evaluate the role of the continents and seafloor in controlling global
carbon cycling.
Together, these three chapters represent three different approaches to modeling
climate change throughout Earth’s history, and emphasize the importance of using both
climate models and geochemical proxies to study this fascinating problem.
7
Chapter 2.
Iron micrometeorite oxidation as a proxy for atmospheric CO2 during the Neoarchean
Chapter 2 is published in the Proceedings of the National Academy of Sciences (PNAS) as of January 2020 (https://doi.org/10.1073/pnas.1910698117), with coauthors Don Brownlee and James Kasting. I ran the modeling cases described in the chapter and wrote the analysis. I thank both coauthors for their input developing the calculations, and their contributions to some of the text. 1. Background
Earth’s atmospheric O2 concentration is widely believed to have been low or
negligible prior to ~2.5 Ga, based on a variety of geologic evidence (Holland, 2006),
including multiple sulfur isotopes preserved in sediments (Farquar et al., 2000; Pavlov et
al., 2001; Harman et al., 2018). These constraints on O2 are relevant to the lower
atmosphere—as sediments only interact with the air directly above the ground—but do not
necessarily apply to the upper atmosphere. Tomkins et al. (2016) proposed that ancient
spherical micrometeorites could be used to determine past oxygen concentrations in the
upper atmosphere. They extracted fifty-nine iron-type micrometeorites from Pilbara
limestone, dated at ~2.72 Ga, and used the iron oxides contained within them to estimate
the amount of O2 the micrometeorites would have encountered as they were melted during
entry. These type I cosmic spheres were small Fe-Ni metal meteoroids that totally melted
to form spheres during hypervelocity entry into the atmospheric. Their oxygen content was
obtained solely in the upper atmosphere during a brief period of frictional melting.
Tomkins et al. (2016) concluded that upper atmospheric O2 concentrations must have been
close to the modern value, 21 percent by volume, to create the largely oxidized composition
of the micrometeorites. To explain the discrepancy with inferred low O2 concentrations at
the surface, they suggested that vertical atmospheric mixing in the Neoarchean could have
8
been inhibited by the presence of a stratospheric organic haze, which would have caused a
temperature inversion by absorbing incoming sunlight and, thus, isolated upper
atmospheric O2 from O2 in the lower atmosphere.
In more detail, Tomkins et al. argued that the micrometeorites would have passed
through the upper atmosphere, accelerated by Earth’s gravitational field, and reached
maximum temperature and velocity between 85 and 90 km above the surface (Love &
Brownlee, 1991). They posited that melting and quench-cooling of these sand grain-size
meteorites would have occurred within approximately two seconds between 75 and 90 km,
so this is where oxidation should have occurred. Tomkins et al. argued that the
micrometeorites must have been oxidized when passing through the atmosphere, and not
by later alteration, based on several lines of evidence. First, the existence of a layer of
encasing wüstite (FeO) in some of the micrometeorites supports an extraterrestrial origin
of the spherules, as this iron oxide is not formed in sedimentary environments. Second, the
presence of FeO and magnetite (Fe3O4), and the absence of hematite, in any of the collected
meteorites indicates oxidation is unlikely to have occurred in a sedimentary environment
where water may have been present. Third, the preservation of delicate surface textures
indicates that later chemical alteration or diagenesis probably did not occur, as these
textures would not have survived. Tomkins et al. then developed a mathematical model to
determine the amount of O2 that would need to have been encountered during atmospheric
entry to produce the iron oxides found in the micrometeorites. This included an equilibrium
chemical model coupled to a numerical model of meteorite ablation. Tomkins et al. argued
that CO2 would have been an ineffective oxidant on its own because of its supposedly
9
sluggish rate of reaction, and therefore free O2 would have been needed to oxidize the
meteorites.
To test both this theory of micrometeorite oxidation, we used our own 1-D
photochemical model to create a suite of Archean atmosphere profiles with varying
concentrations of atmospheric CO2. Rather than focus exclusively on upper atmospheric
O2, as Tomkins et al. did, we included O in our calculations as well, in order to assess the
availability of all forms of oxygen. Furthermore, we reevaluated the efficacy of CO2 as an
oxidant, and considered atmosphere-particle interactions to determine likely conditions for
micrometeorite oxidation in the upper Archean atmosphere. We then simulated the effect
of possible CO2 concentrations on surface temperature and pressure, for different values of
pN2, using a 1-D radiative-convective climate model.
2. Micrometeorite oxidation 2.1. Entry physics
Particles with a diameter of less than 1 mm are too small to generate a bow shock
during atmospheric entry (Love & Brownlee, 1991), meaning that micrometeorites smaller
than this do not create the shock waves or gas caps that typically form around larger
particles under these conditions. The interaction between the smaller micrometeorites and
the atmosphere is therefore thought to be simple, as no shock-induced chemical alteration
occurs to the surrounding air during entry. Instead, the micrometeorite is simply slowed
by, and may react with, the gas it encounters. Deceleration is a function of momentum
exchange, and frictional heating during deceleration is what causes the micrometeorite to
briefly melt and quench-crystallize during atmospheric entry. The speed of the
micrometeorite decreases by a factor of 1/e for each particle mass column of gas that the
micrometeorite impacts, assuming all of the gas momentum is transferred to the
10
micrometeorite. A micrometeorite’s speed would decrease to 0.37 and 0.135 of the initial
speed after colliding with gas equal to 1 and 2 particle masses respectively. Faster
micrometeorites would encounter more air while molten because of their greater
momentum.
The temperature of an entering micrometeorite depends on its velocity and on the
ambient gas density, such that
radiatedpower = !
"∗ -!
#𝜌$%&/ ∗ 𝑣' (1)
Here, 𝜌$%& is the ambient air density and 𝑣 is the micrometeorite’s entry velocity. The
melting temperature of Fe and its oxides is approximately 1500°C, so the micrometeorite’s
speed must be above 6 km/s at 80 km and above 9 km/s at 85 km (in the modern Earth’s
atmosphere) to reach the melting point. To take up oxygen and form the mix of magnetite,
wüstite, and metal observed in the Tomkins et al. spherules, the micrometeorites need to
be molten.
Depending on size and particle speed, most micrometeorites should contact one to
two equivalent masses of air in the time that they are traveling fast enough to be heated to
their melting points. A complication is that the micrometeorites are potentially evaporating
some fraction of their mass during entry, and this is strongly dependent on entry angle and
initial micrometeorite size. Evaporative mass loss during entry would likely decrease the
number of Fe atoms that would need to be reacting with oxidants for the micrometeorite to
be fully oxidized. But, newly acquired oxygen may be lost as well, and so the uncertainties
associated with mass loss make it difficult to constrain in our calculations without knowing
more about entry angle and initial size. For this study, we assume that the micrometeorite
11
retains most or all the oxygen it encounters, along with its own initial mass, so oxidizing
the micrometeorites is therefore a function of how much oxygen the meteorite encounters
in the upper atmosphere while molten.
Tomkins et al. assumed that the altitude region within which the micrometeorites
were molten would be the same in the Archean as it is in the modern atmosphere (between
~75 and 90 km, as Love & Brownlee (1991) specified). But the deceleration—and
frictional heating—of a micrometeorite depends on the number of particles it encounters
during entry. Thus, the window in which oxidation occurs is not located at a fixed altitude
range, but rather at a fixed pressure range. Pressure in our model Archean atmosphere falls
off more rapidly than in the modern atmosphere (see Appendix 1, Fig. A1b) because of the
cooler, ozone-poor stratosphere, along with the higher mean molecular weight caused by
increased CO2. Thus, the ‘oxidation altitude range’ of 75 to 90 km for the modern
atmosphere should really be redefined as an ‘oxidation pressure range’ from ~2.3´10-5 bar
to ~1.6´10-7 bar. The corresponding altitude of oxidation therefore varies with the CO2
concentration and is generally lower than assumed by Tomkins et al. (Table 2.1).
Table 2.1: Approximate altitude of lower and upper bounds of oxidation pressure range for atmospheres with various CO2 mixing ratios. Altitude in kilometers. CO2 (%) Modern 1% 10% 20% 25% 30% 35% 50%
Lower 75 59 55 54 52 51 50 46
Upper 90 72 70 67 65 64 62 58
12
2.2. Micrometeorite oxidation chemistry In a Neoarchean atmosphere with a primarily N2-CO2 composition, upper
atmospheric chemistry would be dominated by the reaction:
CO# + ℎ𝑣 → CO + O (2)
Direct recombination of CO with O is spin-forbidden, and therefore slow, so O primarily
recombines with itself to form O2:
O + O +M → O# +M (3)
In terms of micrometeorite oxidation, it should not matter much whether oxygen
was present as O or O2, as the high temperature of the molten iron micrometeorites should
allow all of the gaseous species interacting with them to equilibrate with each other on
short time scales.
Tomkins et al. only considered O2 as a potential oxidant, assuming that the
concentration of other forms of oxygen would be negligible. Their Fig. 4 shows
calculations by Zahnle et al. (2006) that appear to support this assumption. But our own
photochemical model of the 2.7 Ga atmosphere indicates that both O2 and O would be
present in the upper atmosphere (see Section 3 of this chapter). The Zahnle et al. model
(Zahnle et al., 2006; Zahnle, 1986), which is a derivative of our model, would almost
certainly yield the same result. However, those authors were focused on the lower
atmosphere and simply did not include O in their figure.
It is also possible—even likely—that CO2 itself was an oxidant for the
micrometeorites. Tomkins et al. argued that Fe oxidation by CO2 would be slow, based on
studies of Fe metallurgy at temperatures below 1470 K (Abuluwefa et al., 1997; Bredesen
13
& Kofstad, 1991). But iron micrometeorites (and iron oxide) need to reach at least ~1770
K during entry in order to melt (Love & Brownlee, 1991). Shock tube experiments above
the Fe melting temperature indicate that oxidation rates of Fe to FeO by O2 (Akhmadov et
al., 1988) and CO2 (Giesen et al., 2002) are roughly equal and are up to 3 orders of
magnitude faster than reduction of FeO by CO (Smirnov, 2008). Other possible oxidation
reactions include O2 oxidation of Fe to FeO2 (Smirnov, 2012), and CO2 oxidation of FeO
to FeCO3 (Rollason & Plane, 2000) at similar reaction rates. But it is worth noting that O
can also reduce FeO to form O2 in the upper mesosphere (Self & Plane, 2003), albeit at
much lower temperatures than the micrometeorites would experience during entry. Until
experimental chemical reaction rate data are obtained for the appropriate temperature and
pressure range, we cannot definitively state which oxidation and reduction reactions are
likely to dominate. Nonetheless, the CO2 oxidation pathway indicates the value such
experiments would have for micrometeorite analysis—and, for now, existing data points
to its potential significance for the Tomkins micrometeorites. In short, it suggests that at
least one O atom from each CO2 molecule may contribute to micrometeorite oxidation
during entry. It also suggests that reduction of the newly formed Fe oxides by CO is
unlikely to occur during the short time that the micrometeorite is molten, and that
accumulating an adequate amount of oxidant in the oxidation pressure range is more
important than the ratio of oxidants to reductants.
If so, then one should consider the oxidation potential from O, O2, and CO2 within
the oxidation pressure window. And this, in turn, makes the calculation quite simple, as
nearly all O and O2 in the upper atmosphere is derived from CO2. O atoms are neither
created nor destroyed in the atmosphere above the altitude at which they are removed by
14
rainout of H2O. Thus, from mass balance, it is easy to demonstrate (see Appendix 1) that
if O2 concentrations are low near the surface, and if H2O concentrations are low at the
tropopause, then
𝑓(O) + 2𝑓O# + 𝑓CO# = 𝑐𝑜𝑛𝑠𝑡. = 𝑓CO#( (4)
Here, fi is the volume mixing ratio of species i and 𝑓CO#( is the mixing ratio of CO2 at the
surface. More precisely, 𝑓CO#( is the mixing ratio of CO2 at the base of the stratosphere;
however, the difference between this value and the surface CO2 mixing ratio is negligible.
Thus, in our analysis, the key parameter is the CO2:N2 ratio at the surface. CO2 oxidizes
the micrometeorites, whereas both CO2 and N2 decelerate the meteorites. The CO2:N2 ratio
determines whether or not they are oxidized, independent of particle size.
To make this analysis more concrete, consider the modern atmosphere, within
which we know that incoming metal micrometeorites are partly to fully oxidized to form a
mix of Fe(1-x)O (wüstite), Fe3O4 and sometimes iron metal during entry (Love & Brownlee,
1991). For total oxidation, this requires the addition of 1 to 1.3 O atoms per iron atom. In
the modern atmosphere, this oxidation is accomplished almost exclusively by O2. Assume
for now that the particle remains molten, and hence reactive, only during its encounter with
the first equivalent air mass. Because the atomic mass of Fe (~56 amu) is just under twice
the average atmospheric mass (29.6 for N2-O2-40Ar), the micrometeorite should encounter
about twice as many air molecules as it contains Fe atoms while it is still molten. Of these,
21% are O2 molecules. Therefore, the ratio of O2 molecules encountered to Fe atoms within
the micrometeorite is equal to 2 ´ 0.21 @ 0.4. The ratio of O:Fe is twice that, or 0.8. This
ratio is close enough to the O:Fe ratio required to form the observed oxides (as the
15
micrometeorites contain some varying fraction of Ni instead of Fe), provided that oxidation
is total and nearly 100% efficient. We can express this relationship compactly by writing
- )*#+.*
/ ∙ 2𝑓O# ≅ 0.8 (5)
Now, let us apply this same logic to the Neoarchean atmosphere, assuming that it consists
primarily of N2 and CO2 within the oxidation pressure window (Figs. 2.1 and A1.2 show
that this is approximately true). The mean molecular weight of the atmosphere is then
M-. ≅ 44𝑓CO# + 28(1 − 𝑓CO#) (6)
Here, fCO2 is effectively 𝑓CO#(, the CO2 mixing ratio at the ground. We will use these terms
interchangeably from here on. Using the identical requirement of 0.8 O atoms per Fe atom
for complete oxidation then yields
H56M-.
K 𝑓CO# =H56𝑓CO#
44𝑓CO# + 28(1 −𝑓CO#)K ≥ 0.8
or
M #012!!34"#5012!
N ≥ 0.8 (7)
Solving for fCO2 gives a minimum value of 0.52 needed to oxidize a micrometeorite
encountering one equivalent mass of air during entry.
Suppose now that the meteorite remains molten during its encounter with two
equivalent air masses. The required CO2 mixing ratio is lower, but not exactly by a factor
of 2. The analog to eq. 7 is
H56M-.
K ∙ 2𝑓CO# =H56𝑓CO#
44𝑓CO# + 28(1 −𝑓CO#)K ≥ 0.8
16
or, more simply,
M "012!!34"#5012!
N ≥ 0.8 (8)
Solving for fCO2 in this case yields a value of 0.23. It is slightly less than half the value
predicted from eq. 7 because the mean molecular weight of the atmosphere is lower,
causing the micrometeorite to encounter more air molecules as it decelerates.
Similar calculations can be performed for the case when CO2 is not considered to
be an oxidant for Fe. We do this here because it remains unclear which assumption is
actually correct. In this case, CO2 and N2 would still be the dominant constituents in the
oxidation pressure range, so Mat remains unchanged, but the fraction of oxidant is fOoxy º
(fO + 2 fO2), rather than fCO2. Thus, if the particle remains molten during encounter with
one equivalent air mass, we can write
H56M-.
K 𝑓O678 =𝑓O678 H56
44𝑓CO# + 28(1 −𝑓CO#)K ≥ 0.8
or
M#∗02$%&!3"#012!
N ≥ 0.8 (9)
The value of fOoxy needed to oxidize the micrometeorites still depends on the mixing ratio
of CO2, but the relationship is more complicated, as O and O2 are both produced, directly
or indirectly, from CO2 photolysis. Thus, a photochemical model is needed to determine
this relationship. We have performed such modeling (see Section 3 of this chapter) and,
not surprisingly, the required atmospheric CO2 mixing ratios for micrometeorite oxidation
using only O and O2 are much higher. For a 2-air-mass oxidation event, the required value
of fOoxy in eq. 9 is cut exactly in half, as the mean molecular mass does not change. But, as
17
we shall see, attaining even this lower value of fOoxy is difficult to achieve in a realistic
Archean atmosphere.
3. Photochemical calculations 3.1. Methods
Photochemical model calculations are required if CO2 is not included as a possible
oxidant for Fe micrometeorites. For our calculations, we used a version of our 1-D
photochemical model developed for high-CO2, low-O2 atmospheres (Harman et al., 2015).
Our model contains 49 chemical species involved in 220 reactions (see Appendix 1 Table
A.1, Supp. Info of Stanton et al., 2018 for the full list). It extends upwards from the Earth’s
surface to 100 km in 1-km thick layers. Vertical mixing is parameterized as a combination
of eddy and molecular diffusion, using a profile appropriate for the modern atmosphere
(Massie & Hunten, 1981). Absorption and scattering of solar radiation were calculated
using a 2-stream algorithm (Toon et al., 1989), assuming a fixed solar zenith angle of 50
degrees. The time-dependent, coupled chemistry/diffusion equations were integrated to
steady state using the (fully implicit) reverse Euler method. We also calculated changes to
the solar UV flux using a parameterization developed by Ribas et al. (2005). Properly
scaling the UV flux is essential for this analysis, as the rate of free oxygen production via
CO2 photolysis depends on this parameter. At 2.7 Ga, the Sun would have been only ~81%
as bright as today in the visible (Gough, 1981; Kasting, 2010), but ~50% brighter than
today at far-UV wavelengths (< 1750 Å), as main sequence stars like our Sun tend to emit
greater amounts of UV radiation earlier in their lifetimes because of their higher rotation
rates and increased magnetic activity (Ribas et al., 2005).
Calculations were performed for a 1-bar, CO2-N2 atmosphere with 1-50% CO2,
along with low concentrations of methane (see below). We should note that we are solving
18
minor-constituent diffusion equations for major species, which introduces some error in
the ratio of CO2:CO:O at high altitudes in the model atmosphere. However, this should
have little effect on our results because, as discussed earlier, it is only the sum of CO2, O2,
and O that matters, as iron reduction by CO is slow. Furthermore, the largest errors in these
ratios occur close to the top of the model atmosphere, well above the altitude range at which
most meteorite oxidation occurs.
We assumed a simplified temperature structure that decreases from 285 K at the
surface (actually, at 0.5 km) to 175 K at 9.5 km and then remains constant above that height
(Fig. A1.1a). This is consistent with predictions from 1-D climate models (e.g., Kasting,
1984), which suggest that the temperature profile of an ozone-free atmosphere should
follow a moist adiabat from the surface up to the tropopause and then become roughly
isothermal above that altitude. We have not attempted to keep the surface temperature
consistent with the assumed CO2 concentration and solar flux in these calculations,
reasoning that upper atmospheric composition should be relatively insensitive to this
parameter. We examine the implications of atmospheric composition and photochemistry
on surface temperature in our climate calculations (see Section 4 of this chapter).
An upward CH4 flux of 3.0´109 molecules cm-2s-1 was assumed for our
photochemical calculations. This is about 3% of the present CH4 flux, and well below the
estimated CH4 flux during the Archean (Kharecha et al., 2005). Unlike Tomkins et al.
(2016), we do not rely on a stratospheric temperature inversion to help build up upper
atmospheric O2, and so we avoid the regime in which fCH4/fCO2 > 0.1 and in which organic
haze may form (Haqq-Misra et al., 2008). The actual amount of CH4 present should have
19
little effect on micrometeorite oxidation; however, it does have implications for climate at
that time, so we return to this issue in Section 4 of this chapter.
3.2. Results
Vertical profiles of major atmospheric constituents for our 25% CO2 case are shown
in Fig. 2.1. Key reducing and oxidizing species in the upper atmosphere at different CO2
concentrations are shown in Fig. A1.2 in Appendix 1. Both O and O2 are present within
the micrometeorite oxidation pressure range, with O dominating in the upper part of this
region and O2 in the lower part. In all these simulations, the sum of fO + 2fO2 (fOoxy) is
much less than fCO2. That is because virtually all the O and O2 is coming from CO2, and
because CO2 itself is relatively resistant to photolysis (CO2 photolyzes only below ~200
nm, where the solar UV flux is relatively low). Accumulating large amounts of O and O2
in the stratosphere and above would require unrealistically low eddy mixing. Tomkins et
al. (5) argued that such low mixing might result from a stratospheric temperature inversion
caused by the presence of organic haze. But the eddy diffusion profile used here already
accounts for this phenomenon, as it was derived for the modern stratosphere which has a
temperature inversion caused by ozone, so our calculations may already overestimate upper
atmospheric oxygen.
Even with CO2 concentrations as high as 50%, the fraction of the atmosphere within
the oxidation pressure range that is composed of O and O2 combined is less than 2% (Fig.
A1.2). Oxidizing the micrometeorites with oxygen alone (eq. 9) requires reaching a value
of fOoxy roughly 10 times higher than this. Doing so would thus require both an extremely
high CO2 concentration and extremely low eddy mixing. It is therefore difficult, or even
impossible, to oxidize the Tomkins et al. micrometeorites using just O and O2, unless O2
20
was abundant throughout the atmosphere. But this possibility is ruled out by geologic data,
including sulfur isotope studies, as mentioned earlier. It is much more likely that the
micrometeorites were oxidized by CO2, in which case the limits on fCO2 derived in Sect.
2.2 remain applicable.
Figure 2.1. Vertical profiles of major constituents mixing ratios for fCO2 = 25%, close to our minimum estimated value. The pressure range for micrometeorite oxidation is indicated by the shaded yellow region. The modern O2 mixing ratio is shown by the vertical dotted line. Atmosphere also contains 0.8 bar N2.
4. Discussion 4.1. Constraints on pCO2 from Neoarchean climate
The high atmospheric CO2 concentrations required to oxidize the micrometeorites
can be compared with CO2 levels required to explain the climate of the Neoarchean Earth.
Ojakangas et al. (2014) have reported diamictites—unsorted terrigenous conglomerate or
21
breccia with a fine matrix, which are commonly interpreted as glacial in origin—dated at
2.7 Ga in the Dharwar Supergroup in India. This is approximately the same age as the
micrometeorites (2.721 ± 0.004 Ga) analyzed by Tomkins et al. (2016). Even more
convincing evidence for glaciation is found in 2.9 Ga rocks from the Pongola Supergroup
in southern Africa (Young et al., 1998), which include diamictites with geochemical
weathering patterns indicative of continental glaciers. Together, these observations suggest
that the climate of the Neoarchean was not too different from today. In a long-term sense,
Earth’s climate is considered to be glacial today because ice caps exist at both poles.
Approximate constraints on global mean surface temperature, TS, during glacial
periods have been estimated by Kasting (1987), and we follow the same approach here.
The modern value of TS is ~288 K. Polar ice caps were absent during the early Cenozoic
and preceding Mesozoic eras. The Antarctic ice cap started to grow about 35 million years
ago, at which time TS was about 5 degrees warmer than today, or 293 K, based on oxygen
isotopes in deep sea carbonate cores. This suggests that 293 K is a reasonable upper limit
for continental-scale glaciation. The argument is not iron-clad, because changes in land-
sea distributions—in particular, the opening of the Drake passage at about this same time—
could also have helped trigger Antarctic glaciation. But we will use this as a reasonable
guess at the upper limit on TS at 2.7 Ga. At the same time, we can take 0oC, or 273 K, as a
reasonable lower limit on TS, as climate models predict that Earth’s climate would go into
a Snowball state if global mean temperatures were to drop much below this value. Climate
theory (Walker et al., 1981) then predicts that silicate weathering would slow, and
atmospheric CO2 would build up if this were the case.
22
We used a 1-D climate model (described in more detail in Chapter 3) to study the
effects of high atmospheric CO2 levels on Neoarchean climate. To do so, we needed to first
establish a relationship between fCO2 (the CO2 mixing ratio) and surface pressure. This
relationship is nonlinear because the total atmospheric pressure changes as fCO2 increases,
given a fixed amount of N2. The required relationship (derived in Section 1.2 of Appendix
1) is
𝑝CO#: =𝑝N#: ∙ -
""#;/ ∙ - 012!
!<012!/ (10)
Here, 𝑝CO#: and 𝑝N#: are the ‘partial pressures’ of CO2 and N2. The term ‘partial pressure’
is used loosely here, as these are not true partial pressures. Rather, they represent the
surface pressure that would be exerted by that amount of gas were it to exist by itself in
Earth’s atmosphere. Lighter gases cause heavier ones to diffuse away from Earth’s surface,
whereas heavier gases cause lighter ones to diffuse towards it; hence, the actual partial
pressure of a gas can be different from its pressure in isolation, seemingly contrary to
Dalton’s Law. The advantage of defining these terms in this way is that surface pressure,
PS, is then given by
P= = 𝑝CO#: + 𝑝N#: (11)
With these relationships in hand, we used the 1-D climate model to calculate mean
surface temperature as a function of fCO2. The results are shown in Fig. 2.2. The assumed
solar luminosity was ~0.81 times present at 2.7 Ga, following Gough (1981). The top panel
shows TS versus fCO2 for different amounts of N2 and CH4. CH4 is a greenhouse gas which
is scarce enough to have little effect on surface pressure, but abundant enough (1000 ppmv
23
in these calculations) to raise TS by 8–10 degrees. This CH4 concentration is an
approximate upper limit derived from Archean ecosystem modeling (Kharecha et al.,
2005). The micrometeorite oxidation constraints are on fCO2, not pCO2, and are displayed
as vertical dotted lines in Fig. 2.2.
The results from the climate model show that if 𝑝N#: were the same as today (~0.8
bar), the climate at 2.7 Ga would have been warm—300 K or more—even if fCO2 was
equal to the minimum value, ~0.23, estimated from 2-air-mass oxidation. For fCO2 = 0.52,
the minimum value for 1-air-mass oxidation, then TS ³ 320 K. Either of these global mean
surface temperatures would almost certainly have precluded polar glaciation.
The results are more promising for a potentially glacial climate if 𝑝N#: was half its
present value, or 0.4 bar. For fCO2 = 0.23, the predicted TS for the no-methane case is ~290
K, which is within the glacial ‘window’. The high-methane case is several degrees warmer
and is outside our nominal window. However, given the uncertainties in estimating this
window, along with the uncertainties in age dating of the micrometeorites and the
glaciations, this result also seems plausible.
All of this suggests that if the micrometeorite oxidation story—with CO2
facilitating Fe micrometeorite oxidation by up to 2 air masses during entry—is correct, pN2
must have been appreciably lower than today back at 2.7 Ga. The history of pN2 in Earth’s
atmosphere is still an area of active debate, and some theoreticians have argued the
opposite. Goldblatt et al. (2009) argued that if the abundance of N2 had been doubled during
the Archean, it could have boosted greenhouse warming by ~4.4 K through collisional
effects with CO2 (a concept we explore for Phanerozoic climate in Chapter 3, though by
24
Figure 2.2. Results from our 1-D climate model: (a) Surface temperature as a function of fCO2, for atmospheres with 0.8 bar (purple) and 0.4 bar (green) of N2. Solid curves are for zero CH4; dashed curves are for 1000 ppm CH4. Blue shaded region denotes sub-freezing global mean surface temperatures; orange shaded region indicates a global mean surface temperature too high to facilitate glaciation (see text). The 1 and 2 air mass oxidation lines represent the fCO2 needed for oxidation to occur if the micrometeorite reacts with these amounts of air during entry. (b) Surface pressure versus fCO2 for a N2-CO2 atmosphere with 0.8 bar N2 (purple) and 0.4 bar N2 (green) N2, as calculated from eq. 11 in the text.
(a)
(b)
25
CO2 interacting with O2 rather than N2). They suggested that N2 would have been higher
in Earth’s early history and was slowly sequestered in the mantle due to biological
drawdown, based on N2-Ar ratios. However, more recent authors have provided empirical
support for lower atmospheric N2 in the past. For example, analysis of vesicles in 2.7 Ga
basaltic lavas erupted at sea-level imply PS < 0.5 bar (Marty et al., 2013), and measured
N2/36Ar ratios in fluid inclusions trapped in 3 to 3.5 Ga hydrothermal quartz suggest PS <
1.1 bar, and possibly as low as 0.5 bar ( Som et al., 2016; Avice et al., 2018). These studies
argue that N2 that was initially part of the mantle would have been exsolved and outgassed
over time. All three of these estimates are consistent with the low pN2 and PS values derived
here.
4.2. Constraints on pCO2 from paleosols
Archean CO2 levels have also been estimated from paleosols. Driese et al. (2011)
published an estimate of 10–50 PAL CO2 at ~2.7 Ga, based on an analytical technique
developed by Sheldon (2008). 1 PAL CO2 corresponds to 370 ppmv, or 3.7×10-4 bar, in
their model, so their estimate is approximately 0.004–0.02 bar. By comparison, our
minimum estimates of pCO2 are ~0.16 bar for the 0.4-bar pN2 case and ~0.25 bar for the
0.8-bar pN2 case (these are true CO2 partial pressures—hence, no prime on pCO2—
obtained by multiplying fCO2 = 0.23 by the corresponding surface pressure in Fig. S1b).
Our pCO2 estimates are clearly much higher than Sheldon’s estimates. But Sheldon’s
method of analysis can be criticized on several different grounds (Kasting, 2014). Most
importantly, it implicitly assumes that every CO2 molecule that enters the soil will react
with a silicate mineral, which is probably not the case. Hence, his method should provide
only a lower limit on atmospheric pCO2. A more recent analysis of the same paleosols by
26
Kanzaki & Murakami (2015) yields pCO2 values ranging from 0.03 bar to almost 0.4 bar
(Fig. 2.3). The upper end of these new estimates overlaps nicely with the CO2 partial
pressures derived here. So, ~0.2 bar might be considered a “best guess” of atmospheric
pCO2 at the time when the Tomkins et al. micrometeorites fell to Earth. At the very least,
the new pCO2 estimates from micrometeorite oxidation provide support for the higher
Kanzaki & Murakami pCO2 estimates from paleosols as compared to the older estimates
from Driese et al. and Sheldon.
Figure 2.3. pCO2 estimates from paleosols compared to those from climate model calculations (grey shaded region) (Kharecha et al., 2005). Estimates from Sheldon (2008) are shown by the dark squares and solid black line (with error in yellow). The dashed vertical bar is the paleosol estimate at 2.7 Ga from Driese et al. (2011). The downward red arrow is the upper pCO2 limit from cyanobacterial sheath calcification at 1.2 Ga (Kah & Riding, 2007). The vertical bars in blue are the paleosol estimates from Kanzaki & Murakami (2015). The upward pink arrow indicates the pCO2 from our calculations at 2.7 Ga if pN2 was 0.8 bar. Modified from Catling & Kasting (2017, Fig. 11.7).
27
4.3. Other analyses of the Tomkins et al. micrometeorites It is worth comparing our analysis of the Tomkins et al. micrometeorites with two
other recent analyses. We discuss their conclusions, and compare them with our own,
here.
Rimmer et al. (2019) reanalyzed the Tomkins et al. micrometeorites, and
concluded that they require low atmospheric surface pressure (on the order of ~0.3 bar
total) at 2.7 Ga to have been oxidized. According to their analysis, this allows H2O to
penetrate into the upper atmosphere, where it then produces O2 from photodissociation by
UV. Zahnle & Buick (2016) had suggested this previously as a possible oxidation route,
though they noted that it would require a much less effective tropopause cold trap than
exists today. This solution would require that H2O was a major upper atmospheric
constituent in the Rimmer et al. model; however, this does not seem to be the case, based
on their Fig. 2, which depicts water vapor concentrations more or less the same as in the
modern atmosphere and in our model. Instead, the O2 in their model must be coming
from the photolysis of CO2, as that is the only species that contains enough O atoms to
produce it. CO2 photolysis is slower in our model than in the Rimmer et al. model, for
reasons that remain to be determined. Overall, their suggested source of O2 from H2O, as
well as their hypothetical atmosphere compositions for both of the cases they tested, are
unlikely, if not unphysical. That said, while we think there may be problems with the
Rimmer et al. model, or at least with their interpretations, we do agree that lower
atmospheric pressure can help explain the Tomkins et al. data, as it would allow for a
higher ratio of CO2:N2. This is a key factor in the analysis we present here.
28
Lehmer et al. (2020) also revisited the Tomkins et al. micrometeorites, using a
model that incorporated entry physics. Their approach focused on modeling the motion,
evaporation, and kinetic oxidation of a micrometeorite entering the atmosphere,
simulating 15,000 entry scenarios by using a random sampling of parameter distributions
for mass, velocity, and entry angle of incoming micrometeorites. In their model, the
micrometeorites start as pure iron and are oxidized to wüstite or magnetite, as the
Tomkins et al. micrometeorite seem to have been. All their simulations were for a 1-bar
atmosphere with CO2 concentrations between 2% and 85%. Lehmer et al. concluded, as
we did, that CO2 was a more likely oxidant than O2 during the Archean if fO2 was 1% or
lower, as was probably the case. While their range of possible fCO2 was considerably
broader than our own—from 0.06 to at least 0.7, compared to our minimum estimate of
0.23 to 0.52—they agree that CO2 would likely have needed to be a major constituent, in
agreement with recent paleosol estimates (Kanzaki & Murakami, 2015). Lehmer et al.
also suggested a lower overall surface pressure during the late Archean, to keep global
mean surface temperatures consistent with glacial evidence from that time. Thus, their
conclusions are basically similar to ours.
5. Conclusions
To truly solve this problem, experimental data on iron oxidation by CO2, O, and O2
in conditions like those a micrometeorite would experience during entry is needed.
Nonetheless, existing data support the idea that the oxidation of Archean iron
micrometeorites, melted during atmospheric entry, depends primarily on the amount of
CO2 available in the atmosphere. Assuming two-air-mass oxidation, and that CO2 itself is
the primary oxidant, we find that at least ~23% CO2 would be needed to oxidize the
29
Tomkins et al. micrometeorites at 2.7 Ga. This CO2 concentration can be reconciled with
values derived from paleosols, provided that one accepts the higher estimates of Kanzaki
and Murakami (2015). It is most easily reconciled with climate models if pN2 was lower
than it is today, as this would facilitate a global mean surface temperature low enough for
glaciation to occur. A surface pressure of about 0.6 bar, with less than 25% CO2, would
have allowed the Tomkins et al. micrometeorites to be oxidized without conflicting with
the evidence for glaciation at 2.7 Ga. There is thus no need to invoke unusually high
atmospheric O2 concentrations to explain the micrometeorite oxidation. Instead, these
oxidized micrometeorites imply a Neoarchean atmosphere that was rich in CO2 and
somewhat poorer in N2 than today’s atmosphere.
30
Appendix 1: Full Derivations and Model Specifics A1.1. Full derivation of CO2 mass balance, Eq. (4)
Eq. (4) in Chapter 2 shows that, when CO2 is included as an oxidant, the weighted
sum of the mixing ratios of the different oxidants (O, O2, and CO2) is approximately equal
to the mixing ratio of CO2 at the surface,𝑓CO#(. The derivation is as follows. Start from the
continuity equation:
𝜕𝑛%𝜕𝑡 = 𝑃% − 𝑙%𝑛% −
𝜕Φ%
𝑑𝑧 (A1.1)
and the flux equation, neglecting molecular diffusion (as it is not relevant to this problem
until one gets very near the homopause):
Φ% = −𝐾𝑛𝜕𝑓%𝜕𝑧 (A1.2)
Here, t = time, z = altitude, ni = number density of species i (molecules/cm3), Pi = chemical
production rate (molecules/cm3×s), li = chemical loss frequency (s-1), Φ% = flux of species
i, fi º ni/ n = mixing ratio of species i, K = eddy diffusion coefficient (cm2/s), and n = total
number density. All variables are given in CGS units, largely due to the fact that the kinetic
theory of gases (e.g. Chapman et al., 1990; Chapman & Cowling, 1939) was originally
developed using these units. A derivation of eq. (A1.2), including the molecular diffusion
component, can be found in Ch. 15 of Banks & Kockarts (1973). Eq. (A1.2) applies to
solving for a minor constituent diffusing through a more abundant background gas, which
may introduce errors when applied to gases that become major constituents at some
altitudes. That said, this issue has a limited effect on our analysis largely because the
31
altitudes at which our minor constituents become major constituents is limited to the top
of the atmosphere, above the altitudes of interest for micrometeorite oxidation.
If steady state is assumed, then >?'>@
= 0.If one divides by n to convert number
densities into mixing ratios, eq. (A1.1) becomes
𝑃%𝑛 − 𝑙%𝑓% −
1𝑛𝜕Φ%
𝑑𝑧 = 0 (A1.3)
Now let fi = fOtot º fO + 2 fO2 + fCO + 2 fCO2 (where we have neglected other minor O
species, such as H2O). This variable represents the total oxygen mixing ratio. Continuing
to refer to this mixing ratio as fi, for simplicity, we can write
𝑃%𝑛 − 𝑙%𝑓% = 0 (A1.4)
Eq. (A1.4) must be true because O atoms are neither created nor lost above the altitude at
which they are removed via rainout of H2O.Inserting eq. (A1.4) into (A1.3) then yields
𝜕Φ%
𝑑𝑧 = 0 →Φ% = 𝑐𝑜𝑛𝑠𝑡. (A1.5)
But Φ% must be zero at the top of the atmosphere, so we know from eq. (A1.2) that
𝜕𝑓%𝜕𝑧 = 0 →𝑓% = const. (A1.6)
and, therefore
fOtot = fO + 2 fO2 + fCO + 2 fCO2 = const. ≈ 2𝑓CO#( (A1.7)
32
where 𝑓CO#( is the CO2 mixing ratio at the Earth’s surface. Eq. (A1.7) is valid if fO, fO2,
and fCO are all <<f CO2 at the surface; we see this is the case in our photochemical
calculations (see Fig. A1.2, below).
Now, do the same calculation for carbon. If we define the total carbon mixing ratio
as fCtot º fCO + fCO2, then by the same logic as before we must have
fCtot = fCO + fCO2 ≅ 𝑓CO#( (A1.8)
Now subtract eq. (A1.8) from eq. (A1.7) to get
𝑓(O) + 2𝑓O# + 𝑓CO# = 𝑐𝑜𝑛𝑠𝑡. = 𝑓CO#( (A1.9) which is Eq. (4) in the main text of Chapter 2. A1.2: Derivation of Eq. 10
We use ‘pseudo’ partial pressure variables, pCO2¢ and pN2¢ in eq. (10) in Chapter
2. These variables are defined as the surface pressure that would be exerted by a certain
amount of gas if it was present by itself in the atmosphere. This definition sometimes leads
to confusion because it is seemingly contradictory to Dalton’s Law, which says that the
partial pressure of any ideal gas in a mixture of such gases is independent of the partial
pressures of the other gases. Both CO2 and N2 are effectively ideal at terrestrial atmosphere
pressures and temperatures, and yet Dalton’s Law does not hold. The reason is that
Dalton’s Law is derived for gas mixtures that are contained within a fixed-volume
container. The atmosphere is not fixed volume. Instead, the gases are confined by the
planet’s gravitational field. Lighter gases tend to diffuse away from the surface, while
heavier gases diffuse towards it. The gases do work on each other as they interact through
33
diffusion. Adding a lighter gas, e.g. N2, to a heavier gas, e.g., CO2, reduces the partial
pressure of CO2 at the surface because the lighter N2 drags some of the heavier CO2 away
from the planet’s surface, thereby lowering its partial pressure.
In our analysis, though, we need to know the actual CO2 partial pressure at the
surface because that is what is measured, or at least attempted to be measured, from
paleosols. And we need to be able to relate the CO2 volume mixing ratio, fCO2, to the
pseudo partial pressure variable, pCO2¢, as we do in eq. (10). The derivation of eq. (10)
follows.
The surface pressure of an isolated gas is linearly related to its column abundance.
Thus, for a pure CO2 atmosphere
pCO#: = M12! ∙ g (A1.10)
where, M12! = the column mass of CO2, and g = gravity. The column number densities for
CO2 and N2 can then be found from
NA6B12! =
MA6B12!
44 ∙ mC=
pCO#:
44 ∙ mC ∙ g
NA6BD! =
MA6BD!
28 ∙ mC=
pN#:
28 ∙ mC ∙ g
(A1.11)
where mH = mass of a hydrogen atom. These column number densities are conserved when
the two gases are mixed. The CO2 volume mixing ratio, fCO2, is thus
𝑓CO# =NA6B12!
NA6B12! + NA6B
D! (A1.12)
34
Cancelling out the factors of (mC ∙ g) in eq. (A1.11) yields
𝑓CO# =pCO#:
44_pCO#:
44_ + pN#:
28_
=1
1 + H pN#:
pCO#:K ∙ -4428/
(A1.13)
This equation, rearranged to solve for pCO#: , is
pCO#: =pN#: ∙ H4428K ∙ H
𝑓CO#1 − 𝑓CO#
K (10)
which is our eq. (10) in the main text.
35
Figure A1.1. (a) Assumed vertical temperature profile for all simulations. (b) Vertical pressure profiles for various photochemical simulations, calculated using modern eddy diffusion coefficients. The pressure profile for the modern atmosphere is shown for comparison. (c) Vertical eddy diffusion profile used for all photochemical simulations (“modern”, purple). The solid blue curve is the molecular diffusion profile for atomic oxygen, which is calculated from Banks & Kockarts (1973) eqs. 15.28/29.
36
Figure A1.2. Upper atmospheric profiles of CO (red), O (blue), O2 (green), and CO2 (black) for various assumed surface concentrations of CO2. The shaded yellow region marks the micrometeorite oxidation pressure range for each fCO2.
37
Table A1.1: Full list of reactions in the photochemical model used in Chapter 2 and Chapter 3. Reaction Rate constant
(per second) Reference Notes
1) H2O + O(1D) ➜ 2OH 2.62E-10*EXP(65./T(I)) Dunlea et al. (2004)
°symbol means can be found in NIST database
2) H2 + O(1D) ➜ OH + H
1.10E-10 DeMore et al. (1997)
3) H2 + O ➜ OH + H 3.44E-13*((T(I)/298.)**2.67)*EXP(-3160./T(I))
Baulch et al. (1992)
4) H2 + OH ➜ H2O + H 7.7E-12*EXP(-2100./T(I))
Atkinson et al. (2004)°
5) H + O3 ➜ OH + O2 1.4E-10*EXP(-470./T(I)) DeMore et al. (1992)
6) H + O2 + M ➜ HO2 + M
TBDY(4.11E-32,7.51E-11,1.10,0.,TT,DN)
DeMore et al. (1997); Turanyi et al. (2012)
Explain notation: TBDY(K0,KI,N,M,T,DEN)
7) H + HO2 ➜ H2 + O2 5.6E-12 Atkinson et al. (2004)°
8) H + HO2 ➜ H2O + O 2.4E-12 Atkinson et al. (2004)°
9) H + HO2 ➜ OH + OH 7.2E-11 Atkinson et al. (2004)°
10) OH + O ➜ H + O2 2.4E-11*EXP(110./T(I)) Atkinson et al. (2004)°
11) OH + HO2 ➜ H2O + O2
5.00E-11 Srinivasan et al. (2006)
12) OH + O3 ➜ HO2 + O2
1.7E-12*EXP(-940./T(I)) Atkinson et al. (2004)°
13) HO2 + O ➜ OH + O2 2.7E-11*EXP(225./T(I)) Atkinson et al. (2004)°
14) HO2 + O3 ➜ OH + 2O2
1.97E-16*((T(I)/298)**4.57)*EXP(695./T(I))
Atkinson et al. (2004)°
38
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
15) HO2 + HO2 ➜ H2O2 + O2
2.3E-13*EXP(600./T(I))+1.7E-33*EXP(1000./T(I))*DEN(I)
DeMore et al. (1992)
16) H2O2 + OH ➜ HO2 + H2O
2.9E-12*EXP(-160./T(I)) DeMore et al. (1992)
17) O + O + M ➜ O2 + M 5.21E-35*EXP(900./T(I))*DEN(I)
Tsang and Hampson (1986)°
18) O + O2 + M ➜ O3 + M
TBDY(5.70E-34,1.E-10,2.6,0.,TT,DN)
DeMore et al. (1992); Atkinson et al. (2004)°
19) O + O3 ➜ 2O2 8.0E-12*EXP(-2060./T(I))
DeMore et al. (1992)
20) OH + OH ➜ H2O + O 6.2E-14*((T(I)/298)**2.60)*EXP(945./T(I))
Atkinson et al. (2004)°
21) O(1D) + N2 ➜ O(3P) + N2
1.8E-11*EXP(110./T(I)) DeMore et al. (1992)
22) O(1D) + O2 ➜ O(3P) + O2
3.2E-11*EXP(70./T(I)) DeMore et al. (1992)
23) O2 + hv ➜ O(3P) + O(1D)
5.81E-08 Thompson et al. (1963)
Photolysis rates used were taken from the top of the atmosphere (99.5 km altitude)
24) O2 + hv ➜ O(3P) + O(3P)
4.49E-08 Allen and Frederick (1982)
25) H2O + hv ➜ H + OH 5.48E-06 Thompson et al. (1963)
26) O3 + hv ➜ O2 + O(1D)
3.59E-03 WMO (1985)
27) O3 + hv ➜ O2 + O(3P)
6.97E-04 WMO (1985)
39
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
28) H2O2 + hv ➜ 2OH 5.01E-05 DeMore et al. (1985)
29) CO2 + hv ➜ CO + O(3P)
1.33E-09 Shemansky (1972)
30) CO + OH ➜ CO2 + H 1.5E-13*(1.+0.6*PATM) DeMore et al. (1992)
PATM = DEN(I)*1.38E-16*T(I)/1.013E6
31) CO + O + M ➜ CO2 + M
1.7E-33*EXP(-1510./T(I))*DEN(I)
Tsang and Hampson (1986)°
32) H + CO + M ➜ HCO + M
5.29E-34*EXP(-370./T(I))*DEN(I)
Baulch et al. (1994)°
33) H + HCO ➜ H2 + CO 1.50E-10 Baulch et al. (1992)
34) HCO + HCO ➜ H2CO + CO
0. (Assumed)
35) OH + HCO ➜ H2O + CO
1.69E-10 Baulch et al. (1992)
36) O + HCO ➜ H + CO2 5.00E-11 Baulch et al. (1992)
37) O + HCO ➜ OH + CO 5.00E-11 Baulch et al. (1992)
38) H2CO + hv ➜ H2 + CO
4.16E-05 JPL (1983)
39) H2CO + hv ➜ HCO + H
4.11E-05 JPL (1983)
40) HCO + hv ➜ H + CO 1.0E-2 Pinto et al. (1980)
41) H2CO + H ➜ H2 + HCO
2.14E-12*((T(I)/298)**1.62)*EXP(-1090./T(I))
Baulch et al. (1994)
42) CO2 + hv ➜ CO + O(1D)
1.36E-08 Thompson et al. (1963)
43) H + H + M ➜ H2 + M 6.04E-33*((T(I)/298)**1.00)*DEN(I)
Baulch et al. (1992)°
44) HCO + O2 ➜ HO2 + CO
5.20E-12 Atkinson et al. (2001)°
40
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
45) H2CO + OH ➜ H2O + HCO
8.20E-12*EXP(40./T(I)) Atkinson et al. (2001)°
46) H + OH + M ➜ H2O + M
4.38E-30*((T(I)/298)**-2.00)*DEN(I)
Baulch et al. (1992)°
47) OH + OH + M ➜ H2O2 + M
TBDY(6.9E-31,2.50E-11,0.8,0.,TT,DN)
Atkinson et al. (2004)°
48) H2CO + O ➜ HCO + OH
3.4E-11*EXP(-1600./T(I))
DeMore et al. (1992)
49) H2O2 + O ➜ OH + HO2
1.4E-12*EXP(-2000./T(I))
DeMore et al. (1992)
50) HO2 + hv ➜ OH + O 3.04E-04 DeMore et al. (1985)
51) CH4 + hv ➜ 1CH2 + H2
4.41E-06 Mount et al. (1977)
52) CH3OOH + hv ➜ H3CO + OH
4.07E-05 JPL (1983)
53) N2O + hv ➜ N2 + O 5.34E-07 Johnson and Selwyn (1975)
54) HNO2 + hv ➜ NO + OH
1.7E-3 Cox (1974)
55) HNO3 + hv ➜ NO2 + OH
8.59E-05 JPL (1983)
56) NO + hv ➜ N + O 1.84E-06 Cieslik and Nicolet (1973)
57) NO2 + hv ➜ NO + O 5.78E-03 JPL (1983)
58) CH4 + OH ➜ CH3 + H2O
4.16E-13*((T(I)/298.)**2.18)*EXP(-1230./T(I))
Srinivasan et al. (2005)
59) CH4 + O(1D) ➜ CH3 + OH
1.13E-10 DeMore et al. (1994)
60) CH4 + O(1D) ➜ H2CO + H2
7.51E-12 DeMore et al. (1994)
61) 1CH2 + CH4 ➜ 2 CH3
5.9E-11 Böhland et al. (1985b)
62) 1CH2 + O2 ➜ H2CO + O
6.64E-14 Dombrowsky et al. (1992)
41
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
63) 1CH2 + N2 ➜ 3CH2 + N2
8.8E-12 Ashfold et al. (1981)
64) 3CH2 + H2 ➜ CH3 + H
5.E-15 Tsang and Hampson (1986)
65) 3CH2 + CH4 ➜ 2 CH3
7.1E-12*EXP(-5051./T(I))
Böhland et al. (1985a)
66) 3CH2 + O2 ➜ H2CO + O
6.64E-14 Dombrowsky et al. (1992)
67) CH3 + O2 + M ➜ CH3O2 + M
TBDY(4.49E-31,1.79E-12,3.00,1.70,TT,DN)
DeMore et al. (1997)°
68) CH3 + OH ➜ H2CO + H2
2.59E-13*((T(I)/298)**-0.53)*EXP(-5440./T(I))
Dean and Westmoreland (1987)°
69) CH3 + O ➜ H2CO + H
1.1E-10 DeMore et al. (1992)
70) CH3 + O3 ➜ H2CO + HO2
5.4E-12*EXP(-220./T(I)) DeMore et al. (1992)
71) CH3O2 + HO2 ➜ CH3OOH + O2
3.8E-13*EXP(780./T(I)) Atkinson et al. (2001)°
72) CH3O2 + CH3O2 ➜ 2 H3CO + O2
7.40E-13*EXP(-520./T(I))
Atkinson et al. (2001)°
73) CH3O2 + NO ➜ H3CO + NO2
2.8E-12*EXP(285./T(I)) Atkinson et al. (2001)°
74) H3CO + O2 ➜ H2CO + HO2
3.9E-14*EXP(-900./T(I)) DeMore et al. (1997)
75) H3CO + O ➜ H2CO + OH
1.E-14 NBS (1980)
76) H3CO + OH ➜ H2CO + H2O
3.2E-13 NBS (1980)
77) N2O + O(1D) ➜ NO + NO
6.7E-11 DeMore et al. (1997)
78) N2O + O(1D) ➜ N2 + O2
4.9E-11 DeMore et al. (1997)
79) N + O2 ➜ NO + O 1.5E-11*EXP(-3600./T(I))
DeMore et al. (1992)
80) N + O3 ➜ NO + O2 2.01E-16 DeMore et al. (1997)
(maximum value)
42
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
81) N + OH ➜ NO + H 4.70E-11 Baulch et al. (1994)
82) N + NO ➜ N2 + O 2.09E-11*EXP(100./T(I))
DeMore et al. (1997)
83) NO + O3 ➜ NO2 + O2
1.4E-12*EXP(-1310./T(I))
Atkinson et al. (2004)°
84) NO + O + M ➜ NO2 + M
TBDY(9.E-32,3.E-11,1.5,0.,TT,DN)
DeMore et al. (1997)°
85) NO + HO2 ➜ NO2 + OH
3.60E-12*EXP(270./T(I))
Atkinson et al. (2004)°
86) NO + OH + M ➜ HNO2 + M
TBDY(7.52E-31,3.3E-11,2.4,0.3,TT,DN)
Atkinson et al. (2004)°°
87) NO2 + O ➜ NO + O2 5.50E-12*EXP(190./T(I))
Atkinson et al. (2004)°
88) NO2 + OH + M ➜ HNO3 + M
TBDY(3.28E-30,2.7E-11,3.75,0.,TT,DN)
Troe (2012)°
89) NO2 + H ➜ NO + OH 1.47E-10 Su et al. (2002)
90) HNO3 + OH ➜ H2O + NO3
AK0+AK3M/(1.+AK3M/AK2)
DeMore et al. (1992)
Where: AK0 = 7.2E-15*EXP(785./T(I)), AK2 = 4.1E-16*EXP(1440./T(I)), AK3M = 1.9E-33*EXP(725./T(I))*DEN(I)
91) HO2 + NO2 + M ➜ HO2NO2 + M
TBDY(1.8E-31,4.7E-12,3.2,0,TT,DN)
Atkinson et al. (2004)°
92) HO2NO2 + OH ➜ NO2 + H2O + O2
1.90E-12*EXP(270./T(I))
Atkinson et al. (2004)°
93) HO2NO2 + O ➜ NO2 + OH + O2
7.8E-11*EXP(-3400./T(I))
DeMore et al. (1997)
43
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
94) HO2NO2 + M ➜ HO2 + NO2 + M
(k91)/(2.33E-27*EXP(10870./T(I)))
Calculated from reverse reaction #91
95) HO2NO2 + hv ➜ HO2 + NO2
2.44E-04 JPL (1983)
96) CH3OOH + OH ➜ CH3O2 + H2O
1.9E-12*EXP(190./T(I)) Atkinson et al. (2001)°
97) CH3O2 + OH ➜ H3CO + HO2
5E-11
Assumed = to Reaction #11
98) O3 + NO2 ➜ O2 + NO3
1.40E-13*EXP(-2470./T(I))
Atkinson et al. (2004)°
99) NO2 + NO3 ➜ NO + NO2 + O2
4.5E-14*EXP(-1260./T(I))
DeMore et al. (1997)
100) O + NO3 ➜ O2 + NO2
1.00E-11 DeMore et al. (1997)
101) CH3CL + hv ➜ CH3 + CL
1.27E-07 DeMore et al. (1985)
102) NO + NO3 ➜ NO2 + NO2
1.80E-11*EXP(110./T(I))
Atkinson et al. (2004)°
103) OH + NO3 ➜ HO2 + NO2
2.E-11 Atkinson et al. (2004)°
104) CH3CL + OH ➜ CL + H2O
2.4E-12*EXP(-1250./T(I))
Villenave et al. (1997), Herndon et al. (2001), Hsu and DeMore (1994)
105) CL + O3 ➜ CLO + O2 2.80E-11*EXP(-250./T(I))
Atkinson et al. (2007)°
106) CL + H2 ➜ HCL + H 3.90E-11*EXP(-2310./T(I))
Atkinson et al. (2007)°
107) CL + CH4 ➜ HCL + CH3
8.24E-13*((T(I)/298)**2.49)*EXP(-610./T(I))
Bryukov et al. (2002)
44
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
108) CL + CH3CL ➜ CL + HCL
3.2E-11*EXP(-1250./T(I))
Manning and Kurylo (1997), Beichert et al. (1995), (Wallington et al. (1990)
109) CL + H2CO ➜ HCL + HCO
8.20E-11*EXP(-35./T(I)) Atkinson et al. (2001)°
110) CL + H2O2 ➜ HCL + HO2
1.10E-11*EXP(-980./T(I))
Atkinson et al. (2007)°
111) CL + HO2 ➜ HCL + O2
1.80E-11*EXP(170./T(I))
Atkinson et al. (2007)°
112) CL + HO2 ➜ CLO + OH
6.30E-11*EXP(-570./T(I))
Atkinson et al. (2007)°
113) CL + CLONO2 ➜ CL + CL + NO2 (+O)
6.51E-12*EXP(135./T(I))
DeMore et al. (1997)
114) CL + NO + M ➜ NOCL + M
TBDY(9.E-32,1.E-10,1.6,0.,TT,DN)
DeMore et al. (1997)°
115) CL + NO2 + M ➜ CLONO + M
TBDY(1.3E-30,1.E-10,2.,1.,TT,DN)
DeMore et al. (1997)°
116) CL + NOCL ➜ NO + CL2
5.8E-11*EXP(100./T(I)) DeMore et al. (1997)
117) CL + O2 + M ➜ CLO2 + M
TBDY(1.44E-33,1.E-10,3.9,0.,TT,DN)
DeMore et al. (1997); Atkinson et al. (2007)°
118) CL + CLO2 ➜ CL2 + O2
2.3E-10 DeMore et al. (1997)
119) CL + CLO2 ➜ CLO + CLO
1.2E-11 DeMore et al. (1997)
120) CLO + O ➜ CL + O2 2.5E-11*EXP(110./T(I)) Atkinson et al. (2007)°
121) CLO + NO ➜ CL + NO2
6.2E-12*EXP(295./T(I)) Atkinson et al. (2007)°
122) CLO + NO2 + M ➜ CLONO2 + M
TBDY(1.64E-31,7.00E-11,3.4,0.,TT,DN)
Atkinson et al. (2007)°
123) CLO + HO2 ➜ HOCL + O2
4.8E-13*EXP(700./T(I)) DeMore et al. (1997)
45
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
124) CLO + OH ➜ CL + HO2
1.1E-11*EXP(120./T(I)) DeMore et al. (1985)
125) HCL + OH ➜ CL + H2O
1.70E-12*EXP(-230./T(I))
Atkinson et al. (2007)°
126) HOCL + OH ➜ CLO + H2O
3.E-12*EXP(-500./T(I)) DeMore et al. (1997)
127) CLONO2 + OH ➜ CL + HO2 + NO2
1.2E-12*EXP(-330./T(I)) Atkinson et al. (2007)°
128) HCL + O ➜ CL + OH 1.E-11*EXP(-3300./T(I)) DeMore et al. (1997)
129) HOCL + O ➜ CLO + OH
1.70E-13 Atkinson et al. (2007)°
130) CLONO2 + O ➜ CL + O2 + NO2
4.5E-12*EXP(-900./T(I)) Atkinson et al. (2007)°
131) CL2 + OH ➜ HOCL + CL
3.6E-12*EXP(-1200./T(I))
Atkinson et al. (2007)°
132) CL2 + hv ➜ CL + CL 1.82E-03 DeMore et al. (1985)
133) CLO2 + hv ➜ CLO + O
2.71E-03 DeMore et al. (1985)
134) HCL + hv ➜ H + CL 8.01E-07 DeMore et al. (1985)
135) HOCL + hv ➜ OH + CL
2.84E-04 DeMore et al. (1985)
136) NOCL + hv ➜ CL + NO
2.55E-03 DeMore et al. (1985)
137) CLONO + hv ➜ CL + NO2
4.80E-03 DeMore et al. (1985)
138) CLONO2 + hv ➜ CL + NO3
4.00E-04 DeMore et al. (1985)
139) CLO2 + hv ➜ CL + O2
(#117)/(2.43E-25*EXP(2979./T(I)))
DeMore et al. (1997)
140) HO2 + NO3 ➜ HNO3 + O2
1.91E-12 Becker et al. (1992)°
141) CLO + CLO + M ➜ CL2O2 + M
TBDY(2.05E-32,1.00E-11,4.00,0.,TT,DN)
Atkinson et al. (2007)°
142) CL2O2 + hv ➜ CLO2 + CL
3.21E-03 DeMore (1997)
46
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
143) CL2O2 + M ➜ CLO + CLO
(k141)/(3.E-27*EXP(8450./T(I)))
Calculated from reverse reaction #141
144) CLO2 + M ➜ CL + O2
(k117,I)/(5.7E-25*EXP(2500./T(I)))
Calculated from reverse reaction #117
145) CL + NO3 ➜ CLO + NO2
2.40E-11 Atkinson et al. (2007)°
146) CL + HOCL ➜ CL2 + OH
2.5E-12*EXP(-130./T(I)) Cook et al. (1981) and Vogt and Schindler (1993)
147) CLO + NO3 ➜ CLONO + O2
4.E-13 Atkinson et al. (1992)°
148) CLONO + OH ➜ HOCL + NO2
2.4E-12 Ganske et al. (1992) and Ganske et al. (1991)
*Originally no EXP, but on site -> -1250; Previous value was 3.5E-14 which is far off
149) CLO2 + O ➜ CLO + O2
2.4E-12*EXP(-960./T(I)) Gleason et al. (1994)
150) NO2 + O + M ➜ NO3 + M
TBDY(1.31E-31,2.3E-11,1.50,0.24,TT,DN)
Atkinson et al. (2004)°
151) NO3 + hv ➜ NO2 + O 2.25E-02 JPL (1983)
152) NO3 + NO2 + M ➜ N2O5 + M
TBDY3.70E-30,1.9E-12,4.10,0.20,TT,DN)
Atkinson et al. (2004)°
153) N2O5 + hv ➜ NO2 + NO3
3.16E-04 JPL (1983)
154) N2O5 + M ➜ NO2 + NO3 + M
TBDY(1.3E-19,5.7E-14,0.,0.,TT,DN)
NBS (1980)
47
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
155) N2O5 + H2O ➜ 2 HNO3
2.E-19 (Assumed) Tuning parameter.
156) SO + hv ➜ S + O 0. (Assumed)
157)*
SO2 + hv ➜ SO + O 0.7*(8.63E-05) Warneck et al. (1964), Okabe (1971)
158) H2S + hv ➜ HS + H 1.38E-04 Sullivan and Holland (1966)
159) SO + O2 ➜ O + SO2 1.60E-13*EXP(-2280./T(I))
Atkinson et al. (2004)°
160) SO + HO2 ➜ SO2 + OH
2.8E-11 DeMore et al. (1992)
161) SO + O ➜ SO2 6.0E-31*DEN(I) Kasting (1990)
162) SO + OH ➜ SO2 + H 8.6E-11 DeMore et al. (1992)
163) SO2 + OH ➜ HSO3 TBDY(4.62E-31,1.31E-12,3.90,0.70,TT,DN)
Atkinson et al. (2004)°
164) SO2 + O ➜ SO3 3.4E-32*EXP(-1130./T(I))*DEN(I)
Turco et al. (1982)
165) SO3 + H2O ➜ H2SO4 6.00E-15 DeMore et al. (1992)
166) HSO3 + O2 ➜ HO2 + SO3
1.30E-12*EXP(-330./T(I))
Atkinson et al. (2004)°
167) HSO3 + OH ➜ H2O + SO3
1.00E-11 Kasting (1990)
168) HSO3 + H ➜ H2 + SO3
1.00E-11 Kasting (1990)
169) HSO3 + O ➜ OH + SO3
1.00E-11 Kasting (1990)
170) H2S + OH ➜ H2O + HS
6.10E-12*EXP(-80./T(I)) Atkinson et al. (2004)°
171) H2S + H ➜ H2 + HS 3.66E-12*((T(I)/298)**1.94)*EXP(-455./T(I))
Pen et al. (1999)
172) H2S + O ➜ OH + HS 9.22E-12*EXP(-1800./T(I))
DeMore et al. (1997)
173) HS + O ➜ H + SO 1.60E-10 DeMore et al. (1992)
48
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
174) HS + O2 ➜ OH + SO 4.00E-19 DeMore et al. (1992)
175) HS + HO2 ➜ H2S + O2
3.00E-11 McElroy et al. (1980)
176) HS + HS ➜ H2S + S 1.20E-11 Baulch et al. (1976)
177) HS + HCO ➜ H2S + CO
5.00E-11 Kasting (1990)
178) HS + H ➜ H2 + S 1.00E-11 Langford and Oldershaw (1972)
179) HS + S ➜ H + S2 2.2E-11*EXP(120./T(I)) DeMore et al. (1992), Kasting (1990)
180) S + O2 ➜ SO + O 2.10E-12 Atkinson et al. (2004)°
181) S + OH ➜ SO + H 6.60E-11 DeMore et al. (1992)
182) S + HCO ➜ HS + CO 0. (This reaction has been removed)
182a)*
SO2 + hv ➜ S + O2 0.3*(8.63E-05) Warneck et al. (1964), Okabe (1971)
*note about assumed branching ratio
183) S + HO2 ➜ HS + O2 1.5E-11 Kasting (1990)
184) S + HO2 ➜ SO + OH 1.5E-11 Kasting (1990)
185) HS + H2CO ➜ H2S + HCO
1.7E-11*EXP(-800./T(I)) DeMore et al. (1992)
186) SO2 + hv ➜ SO21 9.66E-04 Warneck et al. (1964), Okabe (1971)
49
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
187) SO2 + hv ➜ SO23 8.84E-07 Warneck et al. (1964), Okabe (1971)
188) H2SO4 + hv ➜ SO2 + OH + OH
8.05E-07 Turco et al. (1979)
189) SO3 + hv ➜ SO2 + O 0. (Assumed)
190) SO21 + M ➜ SO23 + M
1.0E-12 Turco et al. (1982)
191) SO21 + M ➜ SO2 + M
1.0E-11 Turco et al. (1982)
192) SO21 + hv ➜ SO23 + hv
1.5E+3 Turco et al. (1982)
193) SO21 + hv ➜ SO2 + hv
2.2E+4 Turco et al. (1982)
194) SO21 + O2 ➜ SO3 + O
1.0E-16 Turco et al. (1982)
195) SO21 + SO2 ➜ SO3 + SO
4.0E-12 Turco et al. (1982)
196) SO23 + M ➜ SO2 + M
1.5E-13 Turco et al. (1982)
197) SO23 + hv ➜ SO2 + hv
1.13E+3 Turco et al. (1982)
198) SO23 + SO2 ➜ SO3 + SO
7.0E-14 Turco et al. (1982)
199) SO + NO2 ➜ SO2 + NO
1.40E-11 Atkinson et al. (2004)°
200) SO + O3 ➜ SO2 + O2 4.50E-12*EXP(-1170./T(I))
Atkinson et al. (2004)°
201) SO2 + HO2 ➜ SO3 + OH
0. (Assumed)
202) HS + O3 ➜ HSO + O2 9.50E-12*EXP(-280./T(I))
Atkinson et al. (2004)°
203) HS + NO2 ➜ HSO + NO
2.90E-11*EXP(240./T(I))
Atkinson et al. (2004)°
204) S + O3 ➜ SO + O2 1.20E-11 DeMore et al. (1990)
205) SO + SO ➜ SO2 + S 8.30E-15 Herron and Huie (1980)
50
Table A1.1 (continued) Reaction Rate constant
(per second) Reference Notes
206) SO3 + SO ➜ SO2 + SO2
2.00E-15 Chung et al. (1975)
207) S + CO2 ➜ SO + CO 1.00E-20 Yung and DeMore (1982)
208) SO + HO2 ➜ HSO + O2
0. (Assumed)
209) SO + HCO ➜ HSO + CO
(k44) Kasting (1990)
210) H + SO ➜ HSO (k6) Kasting (1990)
211) HSO + hv ➜ HS + O 3.04E-04 DeMore et al. (1985)
212) HSO + OH ➜ H2O + SO
(k11) Kasting (1990)
213) HSO + H ➜ HS + OH (k9) Kasting (1990)
214) HSO + H ➜ H2 + SO (k7) Kasting (1990)
215) HSO + HS ➜ H2S + SO
1.00E-12 Kasting (1990)
216) HSO + O ➜ OH + SO (k13) Kasting (1990)
217) HSO + S ➜ HS + SO 1.00E-11 Kasting (1990)
218) N2 + O1D ➜ N2O TBDY(3.5E-37,1.E-10,0.6,0.,TT,DN)
DeMore et al. (1997)
219) N2O + H ➜ NO + NO + OH
5.03e-7*(298./T(I))**2.16*exp(-18701./T(I))
Bozelli et al. (1994)°
220) N2O + NO ➜ NO2 + N2
2.E-14*EXP(-25000./T(I))
NBS (1980)
51
Chapter 3.
The response of the greenhouse effect, and surface temperature, to changes in atmospheric O2 during the
Phanerozoic Chapter 3 is published in the Journal of Geophysical Research: Atmospheres (JGR: Atmospheres) in August 2016 (https://doi.org/10.1002/2016JD025459), with coauthors Amber Young (née Britt), Howard Chen, David Catling, and James Kasting. I conducted the modeling and wrote the analysis; Britt and Chen contributed to some of the early modeling; Kasting assisted with model development. I thank all coauthors for their input on the text. 1. Background
While the Archean greenhouse effect prior to the rise of atmospheric oxygen
depended largely on CO2, with additional warming from gases like CH4, the Phanerozoic
atmosphere has been markedly different. The last 541 million years of climate history have
been intertwined with the installation of O2 as a major atmospheric constituent, shifting the
fraction of the atmosphere made up of greenhouse gases. This chapter (as well as Chapter
4) focuses on greenhouse effect dynamics in the now-oxygenated atmosphere.
The longstanding paradigm of Phanerozoic climate change is that elevated CO2 at
the beginning of the Phanerozoic and intermittently throughout, along with steadily
increasing solar luminosity, were the dominant controls on global temperature. The
Cenomanian age at ~100 to 94 Ma—the earliest age of the Cretaceous period—was marked
by warm global temperatures. By this time, the Sun was approximately 99% as bright as it
is today, meaning that incoming solar radiation was great enough that the Earth could
maintain a habitable global mean surface temperature with much less atmospheric CO2
than was needed earlier in Earth’s history. The breakup of the supercontinent Pangea
resulted in elevated sea levels as mid-ocean ridge area increased, along with increased
atmospheric CO2 due to volcanic outgassing from spreading centers. CO2 rose to above
52
1000 ppm during this thermal maximum (Berner, 2006; Rothman, 2002). The combination
of near-modern-day solar luminosity and high CO2, driven in part by continental
geography, resulted in a warm global climate during the Cenomanian.
Poulsen et al. (2015) argued that changing atmospheric O2 concentrations may also
have been an important driver of climate during the Phanerozoic, along with changing
atmospheric CO2. Specifically, they used the GENESIS 3-D climate model to simulate a
baseline mid-Cretaceous climate with 21% O2, along with 1120 ppmv CO2, and a solar
constant of 0.9943 times the present value. They then looked at the effect of varying the
atmospheric O2 mixing ratio from 10% to 35%, which correspond to the lower and upper
bounds derived from a charcoal O2 proxy (Scott & Glasspool, 2006). They found that the
global mean temperature increased by 2.1 K for the low-O2 calculation and decreased by
2.3 K for the high-O2 calculation. The suggested mechanism was decreased Rayleigh
scattering at low O2 levels (as fewer O2 molecules in the atmosphere would mean less
scattering), leading to a decrease in reflected solar radiation and a corresponding rise in
surface temperature. At higher O2 levels, increased Rayleigh scattering lowered the surface
temperature in their model. Cloud feedbacks were also argued to be important in
determining the climate response to O2 forcing.
To test whether the Poulsen et al. results are robust, we used a version of our own
1-D climate model (Segura et al., 2003a) to perform a similar calculation. The 2003 model
version was used with the RRTM (Rapid Radiative Transfer Model) of Mlawer et al. (1997)
for infrared radiation because this model has been well-vetted for present Earth conditions.
Instead of simulating the mid-Cretaceous, as Poulsen et al. did, we computed the effects of
changing O2 in the modern atmosphere. In our model, these effects are nearly identical to
53
those obtained for mid-Cretaceous conditions. We also used a 1-D photochemical model
to compute self-consistent ozone changes for these changes in O2, as well as related trace
gases such as CH4 and N2O, and we calculated the accompanying changes in surface
temperature caused by the ozone changes.
2. Conservation of nitrogen
Before describing our calculations, we first point out that one must be careful when
adding or subtracting partial pressures and thus changing overall surface pressure. In
Chapter 2, we considered a similar issue when changing pN2 in the Archean atmosphere to
influence total surface pressure. As noted in Section 4.1 of Chapter 2 (and again below),
lighter gases cause heavier ones to diffuse away from Earth’s surface, but heavier gases
cause lighter ones to diffuse towards it. Therefore, one has to be careful when adding or
subtracting O2 from a model atmosphere, because the partial pressure of a gas is not linearly
related to its column abundance when the surface pressure changes.
Poulsen et al. (2015) calculated the change in atmospheric pressure caused by
adding or removing O2 by simply adding or subtracting partial pressures. It can be shown
that this methodology does not conserve the column mass of other atmospheric
constituents, in this case N2 and Ar. The reason is that, in seeming contradiction with
Dalton's Law, the partial pressure of one atmospheric gas depends on the partial pressures
of the other gases in a column of air under the influence of gravity. Adding a lighter gas to
a heavier one increases the atmospheric scale height and causes the heavier gas to spread
vertically, decreasing its partial pressure at the surface. The errors introduced by incorrectly
changing the atmospheric pressure are modest (about 2%) when expressed in terms of total
atmospheric pressure. However, if one calculates the error relative to the change in
54
atmospheric pressure expected at 10% O2 or 35% O2, the error is much higher, at roughly
10%.
Below, we outline a method for changing the O2 volume mixing ratio while
conserving the column masses of N2 and Ar. For simplicity, we adopt modern volume
mixing ratios of the modern Earth scenario, with values, Ci, of 0.21 for O2, 0.78 for N2, and
0.01 for Ar. To begin, we conserve the current ratio of N2 to Ar, RN2Ar = 1/78. We also
need to initialize the atmospheric column mass, which is given by
𝑀EFG$ =
𝑃H𝑔 (1)
Here, PS is surface pressure and g is gravitational acceleration. We next compute the mean
molecular mass as
𝑚 = c𝐶I! ∙ 32 + 𝐶J! ∙ 28 + 𝐶K& ∙ 40f ∙ 𝑚L (2) where mH is the mass of a hydrogen atom. The column mass of each individual species, i,
is then computed from
𝑀EFG% = 𝐶%𝑀EFG
$ -𝑚%
𝑚/ (3) where mi is the species molecular mass. We then get the column number density of each
gas by dividing its column mass density by its molecular mass
55
𝑁EFG% =𝑀EFG%
𝑚% (4)
When we change the O2 mixing ratio, we must recalculate the mixing ratio of N2 and Ar.
These mixing ratios must sum to unity, such that
𝐶I!: + 𝐶J!( + 𝐶K&: = 𝐶I!: + 𝐶J!:(1 + RD#MN) = 1 (5) So,
𝐶J!( =!<O)!!3P*!+,
(6)
𝐶K&: = 𝐶J!RD#MN (7) We use these new mixing ratios to compute a new mean molecular mass from equation (2).
The rest of the procedure requires iterations, as the total column changes. First, we
calculate a new column number density of O2 from 𝑁EFGI! = 𝐶I!𝑁EFG
$ , where 𝑁EFG$ is the total
column number density. Then, add this to 𝑁EFGJ! + 𝑁EFGK& to get a new value for 𝑁EFG$ ′. Now,
recalculate 𝑁EFGI! using the new value of 𝑁EFG$ ′. 𝑁EFG$ changes in each iteration. Using the new
𝑁EFGI! , we can convert it back to column mass using 𝑀EFG
I! = J-$.)!
'#Q/. Add this value to the
column masses of N2 and Ar (which were held constant) to compute a new total column
mass, 𝑀EFG$ ’. Finally, from this, we find the new surface pressure PS using eq. (1).
To ascertain that we have indeed conserved nitrogen with an altered 𝐶I! value, we
check the final column mass of nitrogen after the final iteration using eq. (3), which is equal
to the initial 𝑀EFGJ! at 7787.43 kg/m2. These numerical results are summarized in Table 3.2.
56
Overall, we find a mean 1.6 % discrepancy between our calculated surface pressure values
and those of Poulsen et al. (2015).
Finally, following the methodology of Poulsen et al., we conserve column masses
of CH4 and N2O by increasing their surface mixing ratios for low O2 and decreasing them
at higher O2 (see Table 3.1). Their mixing ratios are inversely proportional to surface
pressure. This does not account for changes in their atmospheric photochemistry as O2
changes, so we also performed a sensitivity study to determine how important these
assumptions about CH4 and N2O might be. An alternative methodology that of Poulsen et
al. is to hold the upward fluxes of these gases constant and allow their surface mixing ratios
to change freely in response to increasing or decreasing O2. When this approach is taken,
CH4 continues to vary inversely with surface pressure, but the N2O mixing ratio actually
increases as O2 increases, and decreases when O2 decreases (Table 3.1, last two columns),
contrary to what Poulsen et al. assumed. The reason for this is that higher atmospheric O2
shields N2O from photolysis and thus lengthens its atmospheric lifetime. But this has a
negligible effect on surface temperature; as O2 is increased from 21% to 35%, surface
Table 3.1: Effect of changing O2 on surface volume mixing ratios of CH4 and N2O with constant column mass or fixed upward flux Species Volume mixing ratio Constant column mass Fixed upward flux 21% O2 10% O2 35% O2 10% O2 35% O2 CH4 1.60×10-6 1.85´10-6 1.29´10-6 1.91´10-6 1.43´10-6 N2O 3.00×10-7 3.47´10-7 2.42×10-7 2.31´10-7 3.42´10-7
57
temperature rises from 289.54 K to 290.86 K with constant upward flux, as opposed
to 290.95 K with constant mixing ratios, so the effect of properly scaling N2O is
only on the order of approximately 0.1 K).
3. Radiative forcing calculations
Using this methodology, we computed surface pressures for two different O2
volume mixing ratios, 0.1 and 0.35. We then did three sets of calculations using the Segura
et al. (2003) radiative-convective climate model. First, we did a control simulation with
𝐶I!= 0.21 and PS = 101.3 kPa. Results are shown in the second row of Table 3.2. 𝑀j is the
mean molecular weight; TS (equal to 288.3 K) is the calculated mean surface temperature,
AP is the planetary or top-of-atmosphere (TOA) albedo, 𝐹RSTU is the outgoing TOA infrared
flux, 𝐹HIVTU is the reflected TOA solar flux, and 𝐹HIVW? is the downward solar flux at the
surface. Because these are converged simulations, the net TOA flux (net IR minus net
solar) is close to zero.
Next, we did two “one-step” calculations with 𝐶I!changed to 0.1 and 0.35. The
advantage of these types of model calculations is that they provide quick indicators of how
upward and downward energy fluxes in the atmosphere are balanced, which more complex
model calculations (such as the simulations we conduct in Section 4 of this chapter) can
quantify in terms of the effect on surface temperature. The goal of our one-step calculations
was to calculate the instantaneous radiative forcing caused by the O2 change (although
doing so requires making several assumptions, as described below). These simulations
should indicate the direction and size of the change in greenhouse effect compared to
Rayleigh scattering.
58
In these calculations, the vertical temperature profile was kept the same as in the
control run, and the model was either stretched or compressed so as to reach the surface.
Each model atmosphere has 100 vertical layers, which are unevenly spaced in log pressure.
The lowest level is at the surface, so as the surface pressure changes, the pressure at each
level also changes. To maintain a self-consistent atmospheric profile, with a convective
troposphere, we first reduced the temperatures below the atmospheric cold trap to the cold
trap value. However, the surface temperature, TS was kept at the control value of 288.3 K.
We then drew moist adiabats up from the surface until they intersected with the temperature
profile. Finally, we recomputed the water vapor mixing ratio at each pressure level from
its saturation vapor pressure and the specified distribution of relative humidity (following
Manabe & Wetherald, 1967), and then computed radiative fluxes.
Results are shown in Table 3.2 and Fig. 3.1. The outgoing solar radiation decreases
by about 0.05 W/m2 in the low-O2 case and increases by just over 0.4 W/m2 in the high-O2
case. These changes are caused by changes in Rayleigh scattering, which decreases with
low O2 and increases with high O2 as discussed by Poulsen et al. But the outgoing infrared
flux changes by a considerably greater amount: it is 3.87 W/m2 higher in the low-O2 case
and 5.91 W/m2 lower in the high-O2 case relative to the control case. These changes in the
outgoing IR flux are the result of increased and decreased pressure broadening of
absorption lines of CO2 and H2O at high and low O2 levels, respectively. Pressure
broadening is a mechanism by which the effectiveness of a greenhouse gas can be increased
due to atmospheric pressure. Briefly, collisions between greenhouse gases like CO2 and
59
Figure 3.1. Plot of flux as a function of altitude for the one-step simulations, with O2 levels of 10% (blue), 21% (black), and 35% (red).
60
Table 3.2: Effect of changing O2 on radiative fluxes for one-step and converged simulations Uncoupled one-step runs with
unchanged O3 Converged runs with O3 change
O2 % 10% 21% 35% 10% 21% 35% Ps (kPa) 87.61 101.3 125.4 87.61 101.3 125.4 𝑴j (g/mol) 28.52 28.96 29.52 28.52 28.96 29.52
Ts (K) 287.90 288.27 288.95 288.74 289.54 290.95 AP 0.2427 0.2429 0.2440 0.2390 0.2405 0.2431
𝐓𝐎𝐀𝑭𝑰𝑹𝒖𝒑
(W/m2) 260.880 257.400 251.100 258.730 258.230 257.350
𝐓𝐎𝐀𝑭𝑺𝑶𝑳𝒖𝒑
(W/m2) 82.509 82.552 82.973 81.274 81.768 82.651
Surf𝑭𝑺𝑶𝑳𝒅𝒏 (W/m2)
255.58 250.53 241.37 252.12 248.22 241.72
TOA Ftot (W/m2)
-3.3895 +0.0450 +5.9221
non-greenhouse gases like N2 and O2 can broaden the greenhouse gas’ absorption bands
for infrared radiation, allowing it to absorb and interact with a broader wavelength range
and thereby enhance its effectiveness as a greenhouse gas (Li et al., 2009). In other words,
there is more greenhouse warming at higher surface pressure and less greenhouse warming
at lower surface pressure, as expected (Goldblatt et al., 2009b). In our model, the change
in the greenhouse effect caused by changing O2 is greater than the change in planetary
albedo, so the net radiative forcing is -3.39 W/m2 at low O2 and +5.92 W/m2 at high O2.
Our results are similar in sign and general magnitude to independent calculations
performed by Goldblatt (2016) using the SMART (Spectral Mapping Atmospheric
Radiative Transfer) line-by-line radiative transfer model. Thus, we would predict our
61
model should warm at higher 𝐶I! and cool at lower 𝐶I!, just the opposite of the results of
the Poulsen et al. calculations.
The changes in outgoing solar radiation—and associated changes in planetary
albedo—for the one-step radiative forcing calculations in Table 3.2 appear much smaller
than those in Poulsen et al. (2016), and it is reasonable to ask why. First, as pointed out by
Goldblatt (2016), Poulsen et al. evidently conflated Rayleigh scattering with Mie scattering
by cloud particles (see further discussion below). Our model assumes clear skies, and so is
not subject to such problems. Even so, the effects of Rayleigh scattering on outgoing solar
radiation and planetary albedo are quite muted in our model. To see why, we ran additional
one-step simulations with the surface albedo set to zero, so that the calculated upward solar
flux (𝐹HIVTU ) and planetary albedo would be exclusively caused by Rayleigh scattering (Table
3.3); by comparison, the surface albedo is set to 0.27 in our standard model, effectively
simulating the increased reflection caused by clouds. In these “Rayleigh scattering only”
simulations, 𝐹HIVTU increased by 3.25 W/m2 as O2 increased from 21% to 35%, and the
planetary albedo increased by 13.73%. So, Rayleigh scattering does make a measurable
difference in planetary albedo when the surface albedo is low. But for a higher surface
albedo—or for a more realistic atmosphere in which the planetary albedo is dominated by
clouds, as is the case on Earth—Rayleigh scattering is at best a second-order effect.
Although Poulsen et al. did not discuss the effects of pressure broadening, their
version of the GENESIS climate model used the radiation code from CCM3 (Kiehl et al.,
1998), which should include this effect. Indeed, one sees an increase in outgoing longwave
radiation in their 10% O2 case, similar to the increase calculated by our model (see their
62
Supp. Fig. 2b). So, the difference in results between the two models likely arise from other
factors, such as the solar radiation code and/or cloud effects.
Table 3.3: Effect of changing O2 on Rayleigh scattering and the corresponding change in planetary albedo for “Rayleigh scattering only” one-step simulations O2 % TOA𝐹HIV
TU (W/m2)
AP from Rayleigh scattering only
D𝐹HIVTU from
21% O2 (W/m2) D AP from 21% O2
10% 21.575 0.0635 - 1.942 - 8.24% 21% 23.517 0.0692 - - 35% 26.768 0.0787 + 3.251 + 13.73%
4. Converged model calculations
Our one-step radiative forcing calculations described above predict that the climate
should warm at high O2 and cool at low O2—just the opposite from that found by Poulsen
et al. We then tested this prediction by running our own climate model to convergence for
the high- and low-O2 cases. Ozone should also change as O2 changes, and we wanted to
calculate the effect this would have on surface temperature. To do so, we used a 1-D
photochemical model from Segura et al. (2003) to calculate ozone profiles for three
different O2 levels: 10%, 21%, and 35%. The CO2 concentration was held constant at 355
ppmv for these calculations, while the concentrations of CH4 and N2O were proportionately
increased or decreased (from their present mixing ratios of 1.6 ppmv and 0.3 ppmv,
respectively) to keep their column masses constant, as Poulsen et al. did. The O3 profile
was allowed to readjust with changing O2, and this had a measurable effect on our results.
63
Next, we coupled the calculated ozone profiles (Fig. 3.2) to the climate model and ran the
climate model to convergence. Results are shown in Tables 3.2 and 3.4 and Figs. 3.3 and
3.4. The 21% O2 control simulation has a marginally different surface temperature from
the one-step calculation, normalized to 289.54 K, because the ozone profile is different
from the one used to generate Table 3.2 (the one-step ozone profile is preset, while the
converged ozone profile is calculated self-consistently for increased accuracy). Table 3.4
shows Ts calculated for 10% O2 and 35% O2 both with and without the calculated changes
in ozone, as well as the difference in Ts from the control simulation.
Figure 3.2. Vertical profile of ozone for the converged simulations.
64
Figure 3.3. Plot of flux as a function of altitude for the converged simulations. Solid lines indicate simulations with changed ozone profiles according to the concentration of O2, while dashed lines indicate simulations with an ozone profile matching that of the control (21%) run (‘unchanged ozone’).
65
Table 3.4: Effect of the changing O2 on various surface parameters for converged runs with and without variable O3 O2 % Ts (K) ∆Ts (K)
from 21% O2 control
AP 𝐹HIVW? (surface)
O3 column depth (molecules/cm2)
10%, unchanged O3
288.31 - 1.23 0.2399 252.91 8.5079 ´ 1018
10% O3 change
288.74 - 0.80 0.2390 252.12 7.7598 ´ 1018
21% (control)
289.54 0.0 0.2405 248.22 8.5079 ´ 1018
35%, unchanged O3
291.43 + 1.89 0.2418 240.75 8.5079 ´ 1018
35% O3 change
290.95 + 1.41 0.2431 241.72 9.8491 ´ 1018
Figure 3.4. Vertical profile of temperature for the converged simulations.
66
By comparing the calculations with and without the ozone change included, one
can estimate the effect of the changing ozone on surface temperature. Somewhat
surprisingly, including the ozone change reduces ∆Ts by a few tenths of a degree in both
the 10% and 35% O2 cases (Fig. 3.4). This may be because the ‘unchanged ozone’
simulations are actually unphysical, as the ozone layer has been either stretched or shrunk
along with the log-pressure grid and is therefore no longer really the same as in the control
run. The Poulsen et al. results could be similarly influenced by ozone. They used a
prescribed seasonally and latitudinally varying modern ozone profile in all of their
simulations, but their vertical grid should also change as the surface pressure changes. It is
possible, therefore, that their ozone profile is technically unphysical and may be
introducing additional minor errors to their temperature calculations.
We concentrate, then, on the two simulations we did in which ozone is modeled
self-consistently. In this case, when O2 is lowered to 10%, Ts decreases by 0.8 K. When O2
is raised to 35%, Ts increases by 1.4 K. The changes are a factor of two smaller and in the
opposite direction from those calculated by Poulsen et al. (the Poulsen model cooled by
2.3 K in the high O2 case and warmed by 2.2 K in the low-O2 case). We conclude that
changes in the Rayleigh scattering optical depth—identified by Poulsen et al. as the main
driver of O2-forced climate change—are likely not as important as those authors thought.
Poulsen et al. combine Rayleigh and Mie scattering in their scaling analysis to result in a
decrease in reflected sunlight in low-O2 conditions. But, as Goldblatt (2016) points out in
his comment, there is no reason that the number density of cloud droplets should scale with
the number density of air molecules, as the relationship is not co-dependent in the
atmosphere. Therefore, the scaling analysis used by Poulsen et al. cannot be justified.
67
Our own results are consistent with the predictions made earlier based on the
instantaneous radiative forcing calculations made in Section 3 of this chapter: when O2
increases, the change in the greenhouse effect caused by increased pressure broadening
outweighs the change in albedo caused by increased Rayleigh scattering; when O2
decreases, the decreased greenhouse effect has a greater impact on temperature than the
decreased Rayleigh scattering. As for Poulsen et al., their results are highly dependent on
cloud feedbacks, as they were careful to point out in their paper. Radiative forcing from
clouds was 15 to 20 W/m2 higher in the Poulsen et al. model than in other models, which
would influence any potential cloud-driven warming (Goldblatt, 2016). Such feedbacks
have been ignored in our 1-D climate model, in which the effects of clouds are instead
parameterized by adopting a high surface albedo. This does not mean that the Poulsen et
al. calculations are necessarily wrong, but it does suggest that their calculations should be
checked with an independent 3-D climate model because cloud parameterizations are so
uncertain.
5. Discussion 5.1. Additional issues with the Poulsen et al. calculations
We now return to the two observations that motivated the Poulsen et al. calculation.
The first is that paleo-CO2 levels were too low to explain the warm climate of the mid-
Cretaceous and early Paleogene, and the second is that O2 levels were significantly lower
in the Cretaceous than they are today.
Consider the CO2 argument first. Following Bice et al. (2006), Poulsen et al. cite a
median CO2 concentration of 1120 ppmv for the Cenomanian stage, ~100 million years
before present (Ma). But the Bice et al. error bars are at least a factor of two, with a quoted
upper limit on CO2 of 2400 ppmv. Meanwhile, they estimate that tropical sea surface
68
temperatures were at least 5°C warmer than today. Global mean surface temperatures, then,
were probably about 10°C warmer than today, assuming that the poles warmed up to 20°C,
as did Cenomanian deep water (Friedrich et al., 2012). According to Bice et al., attaining
these surface temperatures would require at least 4500 ppmv CO2 in their GENESIS 2.0 3-
D climate model. But different climate models have different sensitivities to increases in
CO2. For a middle-of-the-road climate sensitivity of 3 degrees per CO2 doubling (IPCC,
2014), a CO2 concentration of 2400 ppmv (which represents ~3 doublings from the
preindustrial CO2 level of 280 ppmv) would produce about 9°C of warming on the modern
Earth—almost the amount observed during the Cenomanian. Solar luminosity at 100 Ma
was about 1% less than at present, though, which is equivalent to half a CO2 doubling, or
a temperature decrease of 1.5°C. So, we can account for about 7.5°C of warming out of a
desired 10°C. There may indeed be a deficit of CO2 during the Cenomanian but, if so, it is
not that large, and it could be explained in other ways (such as higher climate sensitivity,
or higher CH4), without invoking changes in O2 concentration.
What about the postulated changes in atmospheric O2? Poulsen et al. cite two
references on this subject: Falkowski et al. (2005) and Tappert et al. (2013). Berner was a
coauthor on the Falkowski et al. paper, and they have relied on his geochemical cycle
model (Berner & Canfield, 1989; Berner et al., 2000) for their predicted Cenomanian O2
levels. An updated version of Berner’s model is described in Berner (2006). This model is
driven by the carbon isotope record in carbonates, which can be used to estimate the rate
of burial of organic carbon in sediments. It predicts high atmospheric O2 (~30%) around
300 Ma, following the deposition of the Carboniferous coal beds, and a minimum of O2 (at
~12%) at the beginning of the Jurassic around 200 Ma. O2 then increases steadily during
69
the Jurassic and Cretaceous, reaching approximately 18% by the Cenomanian (94-100 Ma).
So, the Berner model does not offer much support for low O2 during the period of greatest
climatic warmth in the Cenomanian.
Moreover, O2 models driven by carbon isotopes are poorly conditioned, as small
errors in measured 13C/12C ratios can cause considerable changes to the results. For
example, a 1‰ change in d13C (which is small compared to the scatter in the data)
corresponds to a change in the organic carbon burial rate of ~2´1011 mol/yr. When
integrated over 108 years, this corresponds to an O2 amount of ~2´109 mol, or about half
the O2 content of the modern atmosphere. This may explain why an alternative model for
the evolution of atmospheric O2 (Bergman et al., 2004), which is also driven by carbon
isotopes, predicts 30% O2 at 100 Ma. Both models are heavily influenced by the
uncertainties in the data, as well as the particular form of their C-O cycle parameterization.
Therefore, there is actually little geochemical support for the assertion by Poulsen et al. of
low O2 in the Cretaceous, and other models generally predict O2 close to or in excess of its
abundance in the modern atmosphere (Fig. 3.5).
The other paleo-O2 reference cited by Poulsen et al. is Tappert et al. (2013). These
authors estimate O2 levels from d13C in plant resins. The carbon isotopic composition of
these resins is a complicated function of atmospheric O2 and CO2. Higher CO2 corresponds
to lower O2. When the authors use the higher CO2 levels calculated from models or inferred
from pedogenic carbonates, they estimate O2 concentrations of ~11% at 100 Ma (see their
Fig. 10). So, this paper serves as the basis for the low Cenomanian O2 levels preferred by
Poulsen et al. Preserved d13C in amber has been used to study organic carbon fractionation
and oxidation, though generally as a means of estimating past CO2 fractionation and
70
Figure 3.5. Atmospheric O2 estimates for the Phanerozoic from multiple models—shown are trends from Arvidson et al. in orange (2013; MAGic marine biogeochemistry model), Bergman et al. in green (2004; COPSE climate model), Berner in red (2009; GEOCARB model), Berner & Canfield in purple (1989; mathematical model), and Scott & Glasspool in brown (2010; “fire window” charcoal proxy model). The mean (blue dashed) and range (blue shaded) of the Arvidson et al., Bergman et al., and Berner O2 estimates is indicated, as these most closely agree with ice core reconstructions (Stolper et al., 2016). Present-day O2 is shown by the horizontal black line. Of note is that, though some estimates of O2 at 100 Ma are below the modern level, most estimate O2 was higher than 21%, and much higher than the 10% estimate from Poulsen et al. (2015). Figure modified from Wade et al. (2019).
oxidation, rather than as a measure of atmospheric O2 (e.g. Lyons et al., 2019; Royer &
Hren, 2017). It remains to be determined whether the connection between resin d13C and
paleo-O2 from Tappert et al. (2013) is robust.
As one more reason for skepticism, we also wonder whether an O2 level as low as
11% would have been enough to support the large, active dinosaurs and early placental
mammals that lived during the Cenomanian, both of which have considerable oxygen
71
demands. O2 abundance is considered an extrinsic control on gigantism; bigger terrestrial
animals generally have a higher oxygen demand because of their massive size and energy
demands (Beerling, 2017; Sander et al., 2011). The lung-air sac system of birds decrease
weight and increase breathing efficiency, and may have evolved in massive sauropods (i.e.
Brontosaurus, etc.) during the mid-Mesozoic (Lambertz et al., 2018) to maximize energy
efficiency. But the presence of this system in sauropods is not certain and, moreover, other
clades of dinosaur such as theropods (i.e. Tyrannosaurus rex, etc.) likely did not have this
adaptation (Quick & Ruben, 2009). Placental mammals—first evolving between 100 and
65 million years ago (Archibald, 1996)—also have high atmospheric O2 demands, as they
require fairly high ambient oxygen to effectively facilitate this reproductive strategy
(Falkowski et al., 2005). From a purely physiological standpoint, it seems that Cenomanian
O2 levels should have been relatively high.
5.2. Comparison with other 3-D models The results of our analysis suggest that the Poulsen et al. results appear to be
driven largely by cloud feedbacks. This merits checking their results against other
independent 3-D climate models, given the disagreement between our results and theirs
and their deviation from the paradigm established by earlier models (namely Berner’s
GEOCARB model and Bergman et al.’s COPSE model, as discussed in Section 5.1).
Recently, Wade et al. (2019) investigated the effect of changes in atmospheric O2 on
climate using the coupled atmosphere-ocean Hadley Centre Global Environmental Model
version 3 (HadGEM3-AO) and Hadley Centre Coupled Model version 3 (HadCM3-BL)
models. Following up on our work in Payne et al. (2016), Wade et al. compared their
72
model with the findings of Poulsen et al. and with our 1-D simulations, and examined the
effect of O2 on Phanerozoic climate.
Overall, Wade et al. found that pO2 has a minor effect on Cretaceous climate, in
agreement with our calculations described in this chapter. Wade et al. theorized that O2
may have had an appreciable impact on climate earlier in the Phanerozoic (such as during
the Permian, as tropical forests were developing and causing major shifts in atmospheric
CO2 and O2), but by the Cenomanian the effect of O2 on surface temperature was likely
minor. Wade et al. concluded, as we do, that cloud feedbacks drive much of the energy
balance uncertainties and surface temperature changes in Poulsen et al. (2015), but that
cloud feedbacks alone are not enough to explain the discrepancies. Error introduced by
unphysical adjustment of atmospheric ozone or changing the partial pressure of O2, as
discussed in this chapter, may explain the additional inconsistency.
The conclusions of Wade et al. (2019)—and of Goldblatt (2016), who first noted
the GENESIS model’s unusually strong cloud radiative forcings—support our findings
discussed in this chapter, and reinforce the conclusion that the Poulsen et al. results are
flawed. The complexity of the controls on atmospheric temperature and pressure, and the
individuality of climate models in general, make it critical that model comparisons like
this are performed when new climate drivers are proposed. The climate model
intercomparison discussed here also highlights the complexity of cloud-related climate
feedbacks, which are still not well understood.
6. Conclusions
In conclusion, the effect of changes in atmospheric O2 on surface temperature
depends on the type of climate model used to study it. Our 1-D, coupled photochemical-
73
climate model predicts a response to changing O2 levels that is about half as large and
opposite in sign from that predicted by the 3-D climate model of Poulsen et al. The
climatic response of our model is dominated by pressure broadening of absorption lines
of CO2 and H2O, which causes the greenhouse effect to increase at higher atmospheric O2
levels and decrease at lower atmospheric O2. Changes in Rayleigh scattering push surface
temperature in the opposite direction, but this effect is outweighed in our model by the
change in pressure broadening. The Poulsen et al. results appear to be largely driven by
cloud feedbacks, and both our calculations and the model comparisons conducted by
Goldblatt (2016) and by Wade et al. (2019) indicate that their calculations are not robust.
Given the large uncertainties in past levels of both O2 and CO2, we agree with Berner
(2006) that Phanerozoic climate has been driven largely by changes in atmospheric CO2
and solar luminosity, coupled with changes in continental geography.
74
Chapter 4.
Long-term changes in atmospheric CO2 and the carbon cycle throughout the Phanerozoic
1. Background
The major shifts in atmospheric composition over the course of the Phanerozoic
make it essential that we hone our understanding of the sources, sinks, and processes that
control the strength of the greenhouse effect. Changes in CO2 and the C cycle in particular
are a major focus of many models and biogeochemical analyses, and yet there is still work
that needs to be done to refine our knowledge of Phanerozoic climate. In the previous two
chapters, we focused on climate questions for two different times in Earth history, the
Neoarchean and the early Cretaceous. In this chapter, we consider changes over time for
the entire Phanerozoic, especially with respect to the processes that influence atmospheric
CO2 and climate in the last 100 Ma.
The classical view regarding the long-term control mechanism for CO2 relies on
the carbonate-silicate cycle (Berner et al., 1983). In this cycle, CO2 is outgassed from
volcanoes and reacts with rainwater to facilitate the weathering of exposed silicate minerals
on the continents. Runoff delivers the reacted CO2 and silicate weathering products to the
ocean, where most of these nutrients are used to make CaCO3 shells. Some of these shells
are, in turn, sequestered in carbonate sediments as these organisms die and their shells sink
to the seafloor. CO2 is eventually returned to the atmosphere over millions of years as the
seafloor is recycled through plate tectonics. The rate of continental weathering dictates the
drawdown of atmospheric CO2, and is dependent on surface temperature and the amount
of continental runoff (Berner, 2006) and, potentially, on pCO2 directly (Walker et al.,
75
1981). On long time scales, the rate of silicate weathering increases as surface temperature
increases—which can be the result of rising atmospheric CO2—resulting in a self-
regulating, stabilizing climate feedback. This long-term cycle works in conjunction with
short-term cycling of carbon, in which CO2 is removed from the atmosphere by surface-
dwelling photosynthetic organisms, both in the ocean and on the continents, and re-released
via organic decay. Rapid changes in atmospheric CO2 (< 1000 years) are dominated by the
short-term feedback, but on longer timescales (> 105 years) the carbonate-silicate cycle
maintains a stable climate and a quasi-steady state for atmospheric CO2 and marine
inorganic carbon (Royer et al., 2014).
There are, however, a number of complications to this long-term cycle. First, some
authors have argued that the temperature dependence of continental weathering is weak
(Krissansen-Totton & Catling, 2017; Walker, 1993) and that atmospheric CO2 is affected
more strongly by surface weatherability. Second, weathering may also occur on the
seafloor, wherein Ca++ ions are leached out of basalts extruded at midocean ridges and then
deposited in carbonate veins and in seafloor pore space. This process acts as an additional
marine CO2 sink (Caldeira, 1995). Studies have suggested that this process occurs
predominantly within the flanks of midocean ridges, away from the ridge axes, and can
proceed at an appreciable rate in vent systems with temperatures as low as 15°C (Coogan
& Gillis, 2018). As a result, deep-ocean temperatures can greatly influence the reaction
rate. While seafloor weathering plays a relatively minor role compared to continental
weathering today, it may have been much more significant in the global C cycle during
warm periods of Earth history that saw higher deep ocean temperatures, such as the
Mesozoic (Coogan & Dosso, 2015; Coogan & Gillis, 2013) and specifically the Cretaceous
76
around ~100 Ma (as we discussed in Chapter 3). Third, biology introduces additional
complications to the carbonate-silicate weathering cycle, particularly during the
Phanerozoic. During the early Phanerozoic, before the advent of vascular plants,
atmospheric CO2 and soil CO2 should have been closely correlated, as the amount of CO2
stored in soil would have been a function of its degree of contact with the air. But as
vascular land plants evolved and established themselves on the continents, organic decay
and the advent of subsurface respiration in plant root systems would have caused soil pCO2
to become many times greater than atmospheric pCO2 (Mora et al., 1991). These vascular
plants first evolved in the late Ordovician Period (by ~444 Ma) and quickly spread and
diversified. This is evidenced by the abundance of plant fossils with xylem (water-
conducting plant tissue that helps to form self-supporting woody stems) and stomata
(specialized plant cells, commonly found in leaves, which control gas and water exchange
between the plant and atmosphere) found in the ~410 Ma Rhynie Chert formation
(Hetherington & Dolan, 2019; Kidston & Lang, 1921). The expansion of terrestrial
forests—most notably, the vast swamp forests that formed the hallmark coal formations of
the Carboniferous, from ~359 to ~299 Ma—and the subsequent rise in terrestrial carbon
storage is often linked to the decline of CO2 from a major to a minor atmospheric
constituent (Royer, 2013; Berner, 2006; Bergman et al., 2004).
Several different proxies have been used to estimate paleo-atmospheric pCO2 over
the course of the Phanerozoic. Some of the most commonly used proxies include paleosols
(which we discussed in Section 4.2 of Chapter 2, in relation to their estimates and reliability
during the Archean, about 2 billion years earlier than the span of Earth history we discuss
in this chapter) and leaf stomatal density, though there are a multitude of other approaches
77
for estimating ancient pCO2. These proxies predict a general decrease of atmospheric CO2
from high levels (> 1000 ppm) early in the Phanerozoic to the comparatively low level
today (with a pre-industrial concentration of ~280 ppm), with limited periods of increase
at ~200 Ma and ~100 Ma (Royer, 2013). There is general agreement between most of the
established proxies in their projections of Phanerozoic changes in pCO2. That said, all
proxies are subject to their own biases. The age, quality, and availability of samples can
influence the accuracy of pCO2 estimates, and uncertainty generally increases as pCO2
increases (see Fig. 4.1a) and can vary by multiple orders of magnitude (Royer et al., 2014).
Overall, while current proxy data gives us a broad picture of trends in pCO2 over the
Phanerozoic, considerable uncertainty remains concerning the magnitude of episodes of
increased atmospheric CO2, and so proxy data alone cannot give the full picture.
There have been several published efforts to simulate changes in atmospheric CO2
as well. Most of these models do so by parameterizing CO2 burial, weathering, and
degassing fluxes over time, and they generally agree with proxy evidence—though it could
be argued that they just contain enough adjustable parameters to be tuned to broadly
coincide with their contained proxy data. The most widely cited and well-known model is
Berner’s GEOCARB model (Berner, 2006a; Berner, 2006b; Berner & Kothavala, 2001).
The GEOCARB model predicts a decline from high CO2 in the early Phanerozoic to lower
concentrations towards the end of the Devonian, with small-scale fluctuations in
atmospheric CO2 throughout the late Phanerozoic (Fig. 4.1b). This generally matches
estimates from proxy data, although there are some periods of major disagreement in the
Mesozoic and Cenozoic (Royer et al., 2014). However, GEOCARB derives its rate for C
burial using C isotope ratios that may not represent a global average (Saltzman & Edwards,
78
2017), and it does not include seafloor weathering (Krissansen-Totton & Catling, 2017).
As we mentioned in Section 5.1 of Chapter 3, small errors in 13C/12C ratios can cause
sizeable errors when calculating the rate of C burial; most pre-Cretaceous δ13C
measurements are taken from inland sea deposits which are isotopically decoupled from
the open ocean and can exhibit internal isotopic disparities due to geography. Another well-
known model is COPSE (Carbon-Oxygen-Phosphorus-Sulfur-Evolution), which is a
process-driven model rather than an isotope-driven model, and which has also been used
to model pO2 (Bergman et al., 2004; Lenton et al., 2018). COPSE estimates organic C
burial rates from biological activity and therefore tracks C isotopes as oxygen tracers rather
than oxygen forcers. But it also shares much of its C cycle infrastructure with GEOCARB.
Thus, COPSE is ultimately subject to many of the same issues with its handling of the
global C cycle as GEOCARB.
A more recent contribution to modeling the global C cycle is a box model by
Krissansen-Totten & Catling (2017) (K-T&C 2017). These authors developed a 2-box
model of the global C cycle that includes both continental and seafloor weathering. The
model solves a series of equations for partitioning of total C between the atmosphere-ocean
system and seafloor pore space over time. Their model utilizes a statistical (Bayesian)
approach that makes it wholly process-driven without drawing on the same isotope data as
earlier models. They also considered long-term effects on climate sensitivity to pCO2, such
as ice sheet retreat and boreal forest decline. Most climate models focus exclusively on
rapid feedbacks such as changes to water vapor or ice caps, as stipulated by equilibrium or
“Charney” climate sensitivity (Charney et al., 1979), based on the assumption that long-
term feedbacks are too slow to matter. The IPCC estimates of 1.5 to 4.5 K per CO2 doubling
79
Figure 4.1: (top) Error analysis for various atmospheric CO2 proxies. Curved lines represent regression fits (CO2 versus associated error) of several studies for each proxy type. Positive and negative errors computed separately. Figure modified from Royer (2014). (bottom) Phanerozoic history of atmospheric CO2 from GEOCARB and various proxies. Solid lines indicate average, with shaded regions denoting error margin. GEOCARB calculations are shown in black/grey; proxy data shown in blue are represented in collective bins spanning 10 million years each. Figure modified from Royer (2014).
80
follow this reasoning. However, when considering climate on a longer timescale (such as
over millions of years), it becomes essential to consider these slower feedbacks, as they
can make a considerable difference (Knutti et al., 2017). The K-T&C model supports this
hypothesis, as their model predicts a long-term climate sensitivity to pCO2 of 5.6<!.#3!.' K per
CO2 doubling, much higher than the IPCC range. K-T&C calculated an activation energy
(EA) for continental weathering of 20<)3!( kJ/mol (much lower than earlier values of EA),
implying that the silicate weathering rate is less temperature-dependent than previously
thought. They concluded that continental weatherability has increased by a factor of 1.7-
3.3 over the last 100 My, due to continental uplift and changes in continental area.
Close examination of the K-T&C model reveals several assumptions that may have
influenced the results. The treatment of the atmosphere-ocean system as a single box in
their model creates potential problems with their handling of temperature and pH gradients.
First, the lack of a biological pump in their model makes it difficult to accurately compare
their calculated pH to proxy data. The deep ocean is slightly more acidic than the surface
ocean, due to the decay of organics and respiration at depth, but the proxy data used by K-
T&C (2017) are representative of surface ocean pH only (Anagnostou et al., 2016). Second,
the K-T&C model assumes that surface and deep ocean temperatures change at
approximately the same rate. But data and model simulations of modern global warming
indicate that temperatures in the deep ocean rise more quickly than surface ocean
temperatures, due to the amplification of warming at the poles where deepwater forms.
This means that the fixed surface-to-deep ocean temperature gradient in the K_T&C model
likely leads to an underestimate of deep ocean warming, along with the effect this would
have on seafloor precipitation and dissolution. Third, the K-T&C model implies a need for
81
a change in continental weatherability over the last ~100 Ma but offers little explanation
of why such a shift would have occurred. They suggest several possibilities—including
changes in sea level, changes in erosion, and the emergence of angiosperms and fungi at
~100 Ma—but ultimately do not offer a compelling solution.
In this chapter, we explore and expand on recent developments in modeling
Phanerozoic CO2. To explore the influences on the global C cycle, we created a box model
that builds on the K-T&C (2017) framework but gives greater consideration to variables
such as pH and the surface-deep ocean temperature gradient. We then used our model to
revisit the K-T&C calculations for continental weatherability and the temperature
dependence of weathering, and we compare our results to those of Walker et al., Berner,
and K-T&C. We then combine our model projections with new understanding of
geography and uplift during the last ~100 Ma—such as that of MacDonald et al. (2019)—
to discuss large-scale trends in C cycling over that time.
2. Methods 2.1. Model development
Both the K-T&C (2017) model and our box model calculate ocean chemistry
explicitly, with consideration given to both carbonate chemistry and alkalinity. Both
models also include pH- and temperature-dependent kinetics, which have classically been
arbitrarily assigned or considered only in isolation (i.e., Berner, 2008; Berner, 2006;
Caldeira, 1995), to calculate the time evolution of the carbon cycle. This box model
approach is also unique because many parameters—such as continental weatherability,
precipitation and dissolution rates, and the marine temperature gradient—are defined as a
range rather than as a fixed value. Outputs are therefore distributions based on parameter
ranges, and this makes it possible to calculate a best fit and confidence range for outputs in
82
spite of uncertainties in some of the coefficients. Geochemical equations and kinetics drive
the model, with proxy data serving only as points of comparison for the model output.
As mentioned above, K-T&C divided the atmosphere-ocean-seafloor system into two
boxes (Fig. 4.2, a), isolating the seafloor. In our model, we track carbon and alkalinity
fluxes into and out of three boxes (Fig. 4.2, b)—the atmosphere-surface ocean, the deep
ocean, and seafloor pore space—to more realistically study the effect of surface-to-deep
ocean pH and temperature gradients on seafloor precipitation and dissolution. Carbon
enters the atmosphere-surface ocean through carbonate and silicate weathering on the
continents, as well as through volcanic outgassing. Weathering creates an alkalinity flux
into the atmosphere-surface ocean; the ingoing and outgoing alkalinity fluxes are double
any accompanying carbon flux because the uptake or release of one mole of carbon (as
carbonate) is balanced by a cation with a charge of 2+, e.g., Ca2+. C and alkalinity leave
the ocean via carbonate burial on the relatively shallow continental shelves as well as
carbonate deposition in the deep ocean, and by two-way mixing between the deep ocean
and seafloor pore space as seawater circulates. Additional alkalinity is supplied to the pore
space by dissolution of the basalts that make up much of the seafloor, and alkalinity and C
are lost from pore space as carbonates precipitate out. Details on the dynamical equations
and model setup can be found in the Methods section and in the Supplemental Information
section of K-T&C (2017).
C and alkalinity are exchanged at steady fixed rates between the surface and deep
ocean. However, our 3-box model incorporates a pH gradient between the surface and deep
ocean that is parameterized similarly to observations of the modern ocean, where the deep
ocean is approximately half a log unit more acidic than the surface ocean due to the sinking
83
Figure 4.2 (a): Simplified schematic of K-T&C (2017) box model. Yellow arrows represent C fluxes, while green arrows represent alkalinity fluxes. Total C and alkalinity within the atmosphere-surface ocean-deep ocean and the seafloor pore space is determined based on reservoir mass, fluxes, and internal chemistry. Alkalinity fluxes are twice those of C because the uptake/release of one mole of C as carbonate is balanced by a 2+ cation.
Figure 4.2 (b): Simplified schematic of the modified box model used in our calculations. Both the C and alkalinity fluxes are the same as in the K-T&C model, but the deep ocean is a distinct box in our version, allowing for a more realistic calculation of pH and temperature gradients and the effect this has on seafloor C storage. We also modify continental weatherability and its temperature dependence.
84
and decay of organics via the biological pump (Kump et al., 2010). K-T&C did not include
a biological pump in their model, arguing that its relatively fast response time (< ~1000
years) would not have lasting effects on C cycling. But the difference between surface and
deep ocean pH that is in part generated by the biological pump may have an appreciable
effect on the ocean’s saturation state and the precipitation/dissolution of seafloor
carbonates, so we include this in our 3-box model. We also implement a surface-deep ocean
temperature gradient derived from modern 3-D global climate models. K-T&C assumed
that surface ocean and deep ocean temperatures change at approximately the same rate. But
more detailed ocean modeling tells a different story: at high latitudes (specifically the
Arctic), temperatures increase at a faster rate than the global mean temperature. This “polar
amplification” of warming in coupled atmosphere-ocean GCMs causes polar temperatures
to increase 1.5 to 4.5 times faster than global mean warming (Holland & Bitz, 2003). The
magnitude of polar amplification decreases in warmer climates as ice cover decreases. If
one assumes that the temperature of ocean deepwater—which today forms at high
latitudes—always reflects polar temperatures (Poulsen & Zhou, 2013), then the rate of
change for surface and deep ocean temperatures should not be the same. Accurately
representing the surface-to-deep ocean temperature gradient is important for quantifying
modern climate change, and is potentially even more important during warmer periods
when it could trigger major shifts in the balance between seafloor C sources and sinks.
Thus, we parameterize the surface-to-deep ocean temperature gradient based on the
estimates from Holland & Bitz (2003).
85
2.2. Activation energy calculations In addition, we reevaluated the temperature dependence—and EA—of continental
weathering following the same methods as K-T&C (2017). Both Walker et al. (1981) and
K-T&C (2017) expressed the continental silicate weathering flux by scaling it with pCO2
and surface temperature in the modern atmosphere, such that
F=aB = ωF=aBb6c M
pCO#pCO#b6c
Nd
exp H∆T=TeK
(1)
where F=aB is the continental silicate weather flux (in Tmol/year), F=aBb6c is the modern
continental silicate weathering flux, ω is the weatherability, pCO2 is the actual atmospheric
pCO2, pCO2mod is atmospheric pCO2 in the modern atmosphere, α is a coefficient
representing the relationship between atmospheric pCO2 and soil pCO2, ∆T= is the
difference between the global mean surface temperature at a given time and the pre-
industrial global mean temperature, and Te represents the temperature dependence of
continental weathering. In Equation 1, pCO2 is modified by an exponent α, which has a
range of 0.2 to 0.5 in the model based on measurements of soil pH at a given atmospheric
CO2 (Volk, 1987). F=aBb6c is modified by a dimensionless weatherability factor, ω, that
represents the change in continental weatherability in the past when compared to the
modern. Changes in ω serve to represent changes in sea level, geography, lithology, and
biology collectively. Higher weatherability indicates an increase in weatherable material
on the continents (for instance, by exposure of more land due to a drop in sea level, or by
creation of new continent via tectonic activity). ∆T= is the global mean surface temperature
in the past minus the pre-industrial global mean surface temperature. We use a value of
86
~285 K for the pre-industrial global mean temperature in our calculations because this is
the temperature used by K-T&C, and our aim was in part to compare to their results
(however, we note that this is 1-2 K cooler than the actual pre-industrial global mean
surface temperature; see Kump et al., 2010). Te is a parameter that represents the
temperature dependence of continental weathering; this term is broadly defined by the
approximate change in temperature needed to increase or decrease the weathering rate by
a factor of e. Following Walker et al. (1981), K-T&C relate this “e-folding temperature,”
Te, to an effective activation energy (EA) for continental weathering by the relation:
Te ≈
RTf#
EM
(2)
where TS is surface temperature, EA is in J/mol, and R is the universal gas constant (~8.314
J/mol K). The relationship between Te and EA is approximate because the kinetic
temperature dependence and the temperature dependence of runoff are combined in Te, as
in the K-T&C model and previous modeling efforts. EA represents the energy needed for
the breakdown of silicate rocks on the continents. Estimates of the temperature
dependence, Te, have been made for modern silicate weathering using catchment field
studies of feldspars and other silicates (i.e., Velbel, 1993). Velbel estimated that Te is on
the order of ~9 K, corresponding to an EA of ~77 kJ/mol. A high EA (and low Te) indicates
an extremely strong negative feedback for atmospheric CO2. A smaller activation energy
(and high Te) suggests that the continental weathering feedback is not as closely linked to
surface temperature and is a weaker negative feedback. Walker et al. (1981) included a
temperature dependence with an EA equivalent to ~50 kJ/mol (corresponding to a Te of
87
13.7 K), while GEOCARB (Berner, 2006) utilized an EA of ~75 kJ/mol (meaning, Te = ~9
K), based in part on chemical weathering field studies (West et al., 2005). But K-T&C
found that their “best fit” case (meaning, their model simulations that fit most closely to
proxy data without using the proxy data as priors in a statistical analysis) required a much
higher Te of approximately 30 to 40 K, meaning the EA of continental weathering would
only be ~20 kJ/mol in their projections for climate over the last 100 Ma. It is worth noting
that K-T&C obtained an even better fit using a Bayesian statistical analysis that uses proxy
data as priors; while the Bayesian approach allows for more quantitative conclusions, the
approach we use in our calculations in this chapter present a qualitative look at climate
trends in the last 100 Ma. With this in mind, we test a range of values for Te, covering the
range from K-T&C to GEOCARB.
We ran 100 simulations with our 3-box model, with increasingly narrow parameter
ranges for the model’s input parameters (see Tables 1 and 2 from K-T&C (2017)). For each
simulation, our 3-box model was run for 10,000 iterations to reach a stable solution. Each
iteration sampled uniformly within each of the parameter ranges and then plotted the
resulting change over time for outputs such as atmospheric CO2, ocean temperatures, and
carbonate precipitation and dissolution. The results of each simulation were then compared
to the proxy data used in the K-T&C model, binned in 10-million-year intervals. The “best
fit” was determined in the same manner as the K-T&C best fit referenced above—the “best
fit” case encompassed the most proxy data within the 95% confidence interval for all of
the time series model outputs.
As part of our simulations, we ran several cases testing the effects of a stronger pH
gradient and a stronger surface-deep ocean temperature gradient in isolation. We then ran
88
several cases with varied values for the activation energy of continental weathering, with
our pH and temperature gradients, between the low K-T&C estimate and the much higher
Walker et al. estimate. We did the same with varied past weatherability coefficients,
ranging from the lowest estimate from the confidence range in K-T&C (0.3) to a value
equal to modern weatherability (1.0) and increasing linearly with time. From the results of
these simulations, we calculated a new EA using the equations above, following the
methodology of K-T&C, and the median Te from our “best fit” case.
3. Results
We assess the importance of more realistic surface-to-deep ocean pH and
temperature gradients on the global C system by comparing our model, with these changes
only, to the K-T&C (2017) “base case”. In this case, they used the weathering temperature
dependence calculated by Walker et al. (1981), no change in weatherability with time, no
pH gradient, and an equal rate of temperature change in the surface ocean as in the deep
ocean. We compare their base case with the modified pH and temperature regime used for
our 3-box calculations in Figure 4.3, to isolate their effect from changes in continental
weathering.
Deep ocean temperature had the greatest effect on the saturation state of pore space,
as well as seafloor carbonate precipitation and dissolution. The separation of the surface
and deep ocean in our 3-box model, and the inclusion of a surface-deep ocean temperature
gradient, brings all of the deep ocean temperature proxy data within the confidence interval
on its own (Fig. 4.3, B’). The median carbonate dissolution flux at 100 Ma in our 3-box
model is likewise in close agreement with proxy data (Fig. 4.3, C’). The inclusion of a pH
gradient between the surface and deep ocean had only a minor effect on saturation state,
89
2 BOX MODEL 3 BOX MODEL
Fig. 4.3: Carbon cycle model output for K-T&C 2-box model (A, B, C) and our 3-D box model (A’, B’, C’) for the last 100 Ma. Geochemical proxy data (see K-T&C (2017) for the full list of proxy sources) indicated by dots with bars representing the 95% confidence interval for each. Model median outputs shown by solid lines, with 95% confidence intervals indicated by the corresponding shaded region. Panels denote (A and A’) marine saturation state, (B and B’) surface and deep ocean temperatures, and (C and C’) seafloor dissolution and precipitation carbonate fluxes. The surface-deep ocean temperature gradient in the K-T&C model cases was 0.8-1.4 (meaning rate of temperature change in the deep ocean is 0.8 to 1.4 times that in the surface ocean), while our model implemented a gradient of 1.5-4.5 based on GCM simulations. Our pH gradient resulted in a deep ocean pH 0.5 log units lower than in the surface ocean.
90
decreasing it by ~0.2 (Fig. 4.3, A’). (The term ‘saturation state’ refers to the factor by which
the whole ocean is supersaturated with respect to CaCO3, as in the K-T&C model.) The
minor decrease in saturation state reflects a slight decrease in the concentration of available
CO32- in the deep ocean and in pore space, as would be expected in a more acidic marine
environment. But greater deep ocean warming in the past, at least in part as a result of polar
amplification, has a much larger effect on the fluxes of carbonate precipitation and
dissolution, both of which approximately doubled at 100 Ma compared to the original K-
T&C calculation.
Separating the surface and deep ocean in our model does have an appreciable effect
on the balance between seafloor carbonate sinks and sources. However, these changes to
the marine component of the global C system are not enough to create full agreement
between the model and all of the proxy data. In their “best fit” case (Fig. 4.4), K-T&C
(2017) found that a significant decrease in the temperature dependence of continental
weathering—and overall weatherability during the Cretaceous 100 Ma—was needed to fit
the data. K-T&C obtained the best fit model output by assuming a weak temperature
dependence and overall lower past weatherability for silicate weathering on the continents.
An e-folding temperature of 30-40 K, combined with weatherability 40-60% as strong as
today, resulted in the best fit with the geochemical data.
We found, with the addition of a stronger surface-deep ocean temperature gradient
and biological pump, that weaker weatherability and temperature dependence improved
our fit with proxy data (Fig. 4.5), in agreement with K-T&C. Our best fit was obtained with
an e-folding temperature, Te, of 23<)3+ K and Cretaceous weatherability equal to 50-60% of
the modern value. Using Equation 2, this corresponds to an EA for continental weathering
91
of 29<;3; kJ/mol. This suggests a weaker temperature dependence than predicted by both
Walker et al. (~50 kJ/mol) and Berner (~75 kJ/mol), but not as low a value as that proposed
by K-T&C (20<)3!(kJ/mol). It also supports lower continental weatherability during the
Cretaceous, on the order of what K-T&C suggest in their calculations.
Our “best fit,” with these parameters, broadly agrees with all of the K-T&C proxy
data save for one pH datapoint (Figs. 4.4, 4.5). Our best fits from our time series analysis
are generally comparable with the “best fit” from K-T&C, with the exception of our deep
ocean temperature fits, and show a strong fit with atmospheric CO2 and seafloor dissolution
in particular.
It is worth noting is that our 3-box model median estimate for deep ocean temperatures
does include a deep ocean temperature range that is warmer than the surface ocean at 100
Ma. This prediction is admittedly unphysical—however, the confidence interval for deep
ocean temperatures at 100 Ma includes temperatures below that of the surface ocean. The
wide range of high deep ocean temperatures in the past that our 3-box model estimates
emphasizes the variability introduced by a stronger surface-deep ocean temperature
gradient and polar amplification introduce. Utilizing a Bayesian analysis for our 3-box
model surface-deep ocean temperature gradient would help to more tightly constraint the
strength of this gradient in the last 100 Ma. Our results shown in Fig. 4.5 do, however,
predict a significantly warmer deep ocean in the past that would have strongly influenced
seafloor carbonate precipitation and dissolution in the deep ocean.
92
Fig. 4.4: Carbon cycle model output for the 2-box K-T&C (2017) “best fit” approximation. Solid lines show the median output, while corresponding shaded areas show 95% confidence interval. Geochemical proxy data and error bars indicated by dots. Panels denote (A) surface ocean pH, (B), atmospheric CO2 (ppm), (C) carbonate saturation state, (D) surface and deep ocean temperatures (Kelvin), (E) continental silicate weathering and ocean carbonate precipitation fluxes (Tmol/year), and (F) seafloor dissolution and precipitation of carbonate (Tmol/year).
93
` Fig. 4.5: Carbon cycle model output for our 3-box “best fit” model. Solid lines show the median output, while corresponding shaded areas show 95% confidence interval. Geochemical proxy data and error bars indicated by dots. Panels denote (A) surface ocean pH, (B), atmospheric CO2 (ppm), (C) carbonate saturation state, (D) surface and deep ocean temperatures (Kelvin), (E) continental silicate weathering and ocean carbonate precipitation fluxes (Tmol/year), and (F) seafloor dissolution and precipitation of carbonate (Tmol/year).
94
4. Discussion 4.1. Marine controls on C cycling
Most models of the carbonate-silicate cycle and its effect on climate have focused
on continental weathering as the primary means by which atmospheric CO2 is lost. K-T&C
(2017) included seafloor weathering in their model, arguing that this low-temperature CO2
sink would have played an important role in warm climate episodes in Earth’s past. Our
model results support this hypothesis and suggest that ocean temperatures do play a major
role in marine carbonate preservation, both on the shallow shelves and in the deep ocean.
The importance of deep ocean temperatures to the C cycle is in large part because of its
influence on seafloor weathering and carbonate precipitation. The climate during the mid-
Cretaceous, ~100 Ma, was one of the warmest periods in the Phanerozoic, as we discussed
in Chapter 3. Given that seafloor weathering occurs in off-axis hydrothermal vent systems
and that it depends on deep ocean temperatures (Coogan & Gillis, 2013), it stands to reason
that this process could have been even more important 100 million years ago. During
periods like the Cretaceous when surface temperatures (especially at the poles, where
warming is amplified by positive feedbacks like the ice albedo feedback) were much
higher, deepwater formed at the poles would have likely seen a significant increase in
temperature. It is, therefore, important to not only include seafloor weathering but to also
give full consideration to the controls on deep sea temperatures.
Our box model results indicate that the deep ocean would likely have been at least
10 K warmer at 100 Ma than in the modern ocean. As a result, both precipitation and
dissolution of seafloor carbonates would have been significantly faster, as reaction rates
were elevated by higher temperatures. Due to the strong dependence of seafloor weathering
rates on deep ocean temperatures, it is critical that the controls on ocean temperature are
95
modeled carefully. As our simulations suggest, surface warming by atmospheric CO2 and
the strength of the surface-deep ocean temperature gradient dictate the magnitude of
seafloor weathering and deposition. Underestimating the temperature of deepwater
underestimates the capacity of the seafloor CO2 sink.
It is also possible that warmer deep ocean temperatures could have powered
additional C sinks in the distant past. Isson & Planavsky (2018) argue that there may have
been an additional flux of CO2 from the seafloor via reverse weathering during the
Proterozoic (2,500 – 541 Ma). In this process, excess silica in the oceans can form
authigenic phyllosilicate clays with available cations within seafloor pore space. This
reaction consumes carbonates to form these clays, releasing CO2 back into the deep ocean.
This back-reaction, buffered by deep ocean pH, could have counteracted some of the
drawdown by seafloor weathering. However, reverse weathering would likely only have
made a difference before the rise of siliceous marine life—such as diatoms and some
sponges—which rapidly consume excess silica to form shells (Isson & Planavsky, 2018).
During the Proterozoic, considerable amounts of dissolved silica should have been
available for clay formation, as it was not being utilized on a grand scale by complex shelly
organisms. The earliest of these siliceous organisms evolved around the start of the
Phanerozoic, at ~540 Ma (Li et al., 1998), after which time reverse weathering would have
a negligible impact on ocean CO2 fluxes as the needed silica for clay formation was no
longer available due to widespread biological use. So, while we do not consider reverse
weathering in our analysis here, it may be an important process for pre-Phanerozoic models
of the C cycle.
96
4.2. The role of continental processes in the Phanerozoic C cycle
Ocean temperatures and the effect on seafloor weathering are essential for C cycle
modeling, but continental weathering arguably remains the dominant control on
atmospheric CO2. Silicate weathering drives much of the change over time in the
projections by K-T&C (2017), with lower relative weatherability and a weak temperature
dependence causing the high atmospheric CO2 (and resulting hot climate) at 100 Ma.
Similar to the result of K-T&C, our best-fit estimate of the approximate EA for
continental weathering is considerably lower than previous estimates. As we mention in
Section 2, Walker et al. (1981) included a temperature dependence with an EA equivalent
to ~50 kJ/mol, while GEOCARB (Berner, 2006; Berner, 2008) utilized an EA of ~75
kJ/mol. Our estimate of 29<;3; kJ/mol is much lower, but is not as low as the estimate from
K-T&C (~21 kJ/mol). This difference is likely due to the fact that the K-T&C model
underestimated the CO2 sink from seafloor weathering by neglecting polar amplification
of deep ocean temperature. The much larger differences in EA between our estimate and
those of Walker et al. and Berner may be because neither of their C cycle models included
seafloor weathering, and instead focused solely on carbonate-silicate weathering on the
continents as atmospheric CO2 drivers.
A bigger question may be how and why continental weatherability would have been
lower at 100 Ma compared to today. K-T&C (2017) offered a number of potential
mechanisms for changes in the continent’s weatherability in their study, but ultimately did
not have a clear explanation for the increase over the last 100 Ma that their statistical
analysis implied. The feedback between silicate weathering and surface temperature must
97
be negative over long scales to keep Earth’s climate stable—but the processes influencing
the strength of that climate feedback can vary with time and are not fully understood.
Numerous hypotheses have been proposed for why continental weatherability
might have increased since the Cretaceous, and during the Cenozoic (~65 Ma to present)
in particular. Most of these argue that changes in geography and tectonic processes are the
primary mechanisms. For example, the BLAG model estimated continental land area at
100 Ma was only ~80% that of today (Berner et al., 1983), indicating a higher sea level in
the past. K-T&C likewise assume a higher sea level at 100 Ma in their discussion of
changes to continental weatherability, and point out that studies of ocean basin dynamics
(like those of Müller et al., 2008) suggest that sea level then was at least 80 m higher. The
exposure of more continent surface area since the Cretaceous could drive a significant
fraction of the rise in weatherability, as more land area was exposed to be weathered. But,
as K-T&C point out based on their statistical analysis, sea level alone cannot explain the
full change. Raymo & Ruddiman (1992) proposed that the uplift of the Himalayas starting
at ~50 Ma accelerated silicate weathering fluxes, based partly on rapid episodic increase in
marine Sr isotopes. However, Kump & Arthur (1997) pointed out that this was contradicted
by concurrent C isotopes; they modified this hypothesis with a geochemical model and
proposed that the continent collision that uplifted the Himalayas (~50 Ma to today)
increased continental weatherability rather than weathering fluxes. Later studies have
largely supported this idea and built off of it (Kump et al., 2000). For instance, Maher &
Chamberlain (2014) showed that high topography and new rock can increase
weatherability, and Caves et al. (2016) suggested changing continental lithologies as the
result of uplift could be responsible for shifts in weatherability.
98
A more recent analysis by MacDonald et al. (2019) suggested that the main control
on continental weatherability is low-latitude arc volcanism and associated tectonic activity.
MacDonald et al. cite the emergence of the islands of Indonesia (~20 Ma to today) as the
chief mechanism driving Cenozoic cooling. In this hypothesis, low-latitude arc-continent
collisions create new mafic and ultramafic rock, which is highly reactive due to its
abundant cation content. These newly exposed volcanic rocks are rapidly weathered in the
tropics where warm temperatures and high annual rainfall accelerate mechanical and
chemical weathering. MacDonald et al. argue that episodes of tropical arc-continent
collisions throughout the last 100 Ma (namely the formation of Indonesian islands and
uplift of the Himalayas) could have triggered the observed increase in weatherability and
associated cooling of global climate in the recent Cenozoic. They also argue that previous
arc-continent collisions may have triggered earlier Phanerozoic glaciations. For example,
they suggest that the glaciation in the late Ordovician period was caused by the Taconic
orogeny. Likewise, the major Permian-Carboniferous glaciation was, they argue, the result
of high continental weathering when the Laurasia and Gondwana supercontinents collided
during the formation of Pangea and subsequent uplift of the Allegheny and Appalachian
Mountains. Both the Taconic and Allegheny-Appalachian mountain-building events were
continent collisions that occurred largely in the tropics, between 30°N and 30°S latitude,
where high annual precipitation and warm mean temperatures would facilitate accelerated
rates of silicate weathering.
If MacDonald et al. are correct, it is possible that tectonic activity in the tropics—
along with changes in sea level—caused the increase in weatherability indicated by K-
T&C and our own model calculations. Generating new and highly weatherable continental
99
area could at least partially explain the observed shift. This mechanism for changing
weatherability, in conjunction with the temperature-dependent balance between
continental and seafloor weathering, could explain the long-term shifts we see in global
climate during the last 100 Ma.
5. Conclusions
To more fully solve this problem, it would be necessary to add even greater
resolution to our parameterization of surface-deep ocean gradients (like pH and
temperature) and to constrain our model output with an inverse MCMC analysis. But the
work we present in this chapter offers an in-depth look at the processes controlling
Phanerozoic climate, illustrating the complex balance between C sinks and sources that
must be considered when modeling the Earth system. We agree with K-T&C (2017) that
continental weathering is not as strongly temperature-dependent as earlier models like
GEOCARB (Berner, 2008) have suggested, and an increase in continental weatherability
over the last 100 Ma is needed to explain long-term trends in the global C cycle. However,
we find that seafloor weathering is highly temperature-dependent, and continental
weathering is somewhat more so than K-T&C suggested. A Bayesian analysis using our 3-
box model would more quantitatively assess that difference, but the analysis we present in
this chapter offers a qualitative look at climate change in the last 100 Ma, and emphasizes
the particular importance and complexity of the marine component of the C cycle.
Ultimately, global C cycling is controlled by the strength of both continental and seafloor
weathering, the balance between both C sinks, atmosphere and ocean temperatures, and the
release of atmospheric CO2 by volcanoes. In addition, we agree with recent studies positing
that tectonic processes such as tropical arc-continent collisions and sea level fall can
100
account for much of the increase in weatherability over the last 100 Ma that has caused
atmospheric CO2 to become lower and surface temperatures to decline.
101
Chapter 5.
Concluding Remarks
Studying the history of CO2 and O2 in Earth’s atmosphere shows us a complex story
spanning billions of years of change, with many major shifts in composition driven by
evolving life in the oceans and on land. These changes in the atmosphere have dictated
long-term changes in climate throughout Earth’s history, and so it becomes essential to
study both climate and atmospheric chemistry together to create a complete picture. This,
in turn, requires utilizing a diverse array of techniques, tools, and data.
In Chapter 2, we explored a new potential proxy for CO2 in the Archean
atmosphere, at ~2.7 Ga. Using compositional analysis of ancient iron micrometeorites, we
calculated the amount of oxidants needed to have oxidized the micrometeorites as they
passed through Earth’s atmosphere billions of years ago during entry. From that, we
estimated that the Archean atmosphere likely contained a high abundance of CO2, which
both oxidized the micrometeorites and kept the early Earth warm. We also theorized that
lower pN2 in the past may have been likely to explain both the oxidized micrometeorites
and evidence for glaciation in the late Archean. The work in this chapter points to new
directions for future study, including honing data for CO2 and iron oxidation reactions,
further investigation on long-term changes in atmospheric N2, and applying this possible
new proxy to other periods in Earth history (following future micrometeorite collection
and analysis).
In Chapters 3 and 4, we moved forward in time, focusing on atmospheric evolution
after oxygen rose to prominence in the Earth’s atmosphere. In Chapter 3, we focused on a
102
specific period of climate history at ~100 Ma and conduct an in-depth study of the effect
of O2 on surface temperature and the greenhouse effect during the Cretaceous. Our findings
indicated that O2 accentuates the effectiveness of greenhouse gases like CO2, and support
the longstanding paradigm that CO2, weathering, and geography are the dominant controls
on Phanerozoic climate.
In Chapter 4, we broadened our lens to consider long-term climate change, with
consideration to global carbon cycling since the Cretaceous. We revisited earlier models
of CO2 and the carbon cycle over the last ~100 Ma and expanded on recent modeling efforts
to create a realistic box model of continental and marine influences on global carbon. We
found that continental weathering—and specifically, changes in continental weatherability
due to tectonics and uplift—have driven much of the CO2 evolution since the Cretaceous;
this is supplemented by highly temperature-dependent seafloor weathering. Future work in
this area will include continuing to develop the model used in Chapter 4, and expanding it
to examine the whole Phanerozoic. Future work will include adding more adjustable
parameters to the model and including additional proxy data for comparison with the
model.
Overall, these three chapters examine problems in Earth’s atmospheric history in
three different ways. This work emphasizes the importance of geochemical data and
modeling in studying climate change, past and present, and utilizes a variety of tools to do
so. Ultimately, this research and future work on these topics illustrate that there are still a
number of fascinating questions about Earth’s climate evolution, waiting to be explored.
103
Bibliography
Abuluwefa. (1997). Oxidation of low carbon steel in multicomponent gases. Part 1: Reaction mechanisms during isothermal oxidation. https://doi.org/10.1007/s11661-997-0255-7
Akhmadov, U. S., Zaslonko, I. S., & Smirnov, V. N. (1988). Mechanism and kinetics of interaction of Fe, Cr, Mo, and Mn atoms with molecular oxygen. Retrieved from https://www.osti.gov/biblio/6289832
Anagnostou, E., John, E. H., Edgar, K. M., Foster, G. L., Ridgwell, A., Inglis, G. N., et al. (2016). Changing atmospheric CO2 concentration was the primary driver of early Cenozoic climate. Nature, 533(7603), 380–384. https://doi.org/10.1038/nature17423
Archibald, J. D. (1996). Fossil Evidence for a Late Cretaceous Origin of “Hoofed” Mammals. Science, 272(5265), 1150 LP – 1153. https://doi.org/10.1126/science.272.5265.1150
Arvidson, R. S., Mackenzie, F. T., & Guidry, M. W. (2013). Geologic history of seawater: A MAGic approach to carbon chemistry and ocean ventilation. Chemical Geology, 362, 287–304. https://doi.org/https://doi.org/10.1016/j.chemgeo.2013.10.012
Avice, G., Marty, B., Burgess, R., Hofmann, A., Philippot, P., Zahnle, K., & Zakharov, D. (2018). Evolution of atmospheric xenon and other noble gases inferred from Archean to Paleoproterozoic rocks. Geochimica et Cosmochimica Acta, 232, 82–100. https://doi.org/https://doi.org/10.1016/j.gca.2018.04.018
Banks, & Kockarts. (1973). Aeronomy, Part B. Academic Press.
Beerling. (2017). The Emerald Planet. Oxford University Press.
Bekker, A., Kaufman, A. J., Karhu, J. A., & Eriksson, K. A. (2005). Evidence for Paleoproterozoic cap carbonates in North America. Precambrian Research, 137(3), 167–206. https://doi.org/https://doi.org/10.1016/j.precamres.2005.03.009
Bergman, N. M., Lenton, T. M., & Watson, A. J. (2004). COPSE: A new model of biogeochemical cycling over phanerozoic time. American Journal of Science, 304(5), 397–437. https://doi.org/10.2475/ajs.304.5.397
Berner, & Canfield. (1989). A new model for atmospheric oxygen over Phanerozoic time. American Journal of Science, 289(4), 333–361.
Berner, R. A., Lasaga, A. C., & Garrels, R. M. (1983). The carbonate-silicate geochemical cycle and its effect on atmospheric carbon dioxide over the past 100 million years. American Journal of Science. https://doi.org/10.2475/ajs.283.7.641
Berner, R A, Petsch, S. T., Lake, J. A., Beerling, D. J., Popp, B. N., Lane, R. S., et al.
104
(2000). Isotope Fractionation and Atmospheric Oxygen: Implications for Phanerozoic O<sub>2</sub> Evolution. Science, 287(5458), 1630 LP – 1633. https://doi.org/10.1126/science.287.5458.1630
Berner, Robert A. (2006). GEOCARBSULF: A combined model for Phanerozoic atmospheric O2 and CO2. Geochimica et Cosmochimica Acta, 70(23 SPEC. ISS.), 5653–5664. https://doi.org/10.1016/j.gca.2005.11.032
Berner, Robert A. (2006). Inclusion of the Weathering of Volcanic Rocks in the GEOCARBSULF Model. American Journal of Science, 306(5), 295–302. https://doi.org/10.2475/05.2006.01
Berner, Robert A. (2009). Phanerozoic atmospheric oxygen: New results using the GEOCARBSULF model. American Journal of Science , 309(7), 603–606. https://doi.org/10.2475/07.2009.03
Berner, Robert A, & Kothavala, Z. (2001). Geocarb III: A Revised Model of Atmospheric CO2 over Phanerozoic Time. American Journal of Science , 301(2), 182–204. https://doi.org/10.2475/ajs.301.2.182
Bice, K. L., Birgel, D., Meyers, P. A., Dahl, K. A., Hinrichs, K.-U., & Norris, R. D. (2006). A multiple proxy and model study of Cretaceous upper ocean temperatures and atmospheric CO2 concentrations. Paleoceanography, 21(2). https://doi.org/10.1029/2005PA001203
Bredesen, R., & Kofstad, P. (1991). On the oxidation of iron in CO2+CO mixtures: II. Reaction mechanisms during initial oxidation. Oxidation of Metals, 35(1), 107–137. https://doi.org/10.1007/BF00666503
Caldeira, K. (1995). Long-term control of atmospheric carbon dioxide; low-temperature seafloor alteration or terrestrial silicate-rock weathering? American Journal of Science , 295(9), 1077–1114. Retrieved from http://www.ajsonline.org/content/295/9/1077.short
Catling, & Kasting. (2017). Atmospheric Evolution on Inhabited and Lifeless Worlds. Cambridge University Press.
Caves, J. K., Jost, A. B., Lau, K. V., & Maher, K. (2016). Cenozoic carbon cycle imbalances and a variable weathering feedback. Earth and Planetary Science Letters, 450, 152–163. https://doi.org/10.1016/j.epsl.2016.06.035
Chapman, & Cowling. (1939). The mathematical theory of non-uniform gases: an account of the kinetic theory of viscosity, thermal conduction and diffusion in gases. University Press.
Chapman, Cowling, & Burnett. (1990). The mathematical theory of non-uniform gases: an account of the kinetic theory of viscosity, thermal conduction and diffusion in gases. Cambridge University Press.
105
Charney, E., Lee, C.-H., & Rosenfield, J. S. (1979). Calculations of rotatory strengths in chiral chromophores. Sensitivity to structural parameters in dienes. Journal of the American Chemical Society, 101(23), 6802–6804. https://doi.org/10.1021/ja00517a002
Coogan, L A, & Gillis, K. M. (2018). Temperature dependence of chemical exchange during seafloor weathering: Insights from the Troodos ophiolite. Geochimica et Cosmochimica Acta, 243, 24–41. https://doi.org/https://doi.org/10.1016/j.gca.2018.09.025
Coogan, Laurence A., & Dosso, S. E. (2015). Alteration of ocean crust provides a strong temperature dependent feedback on the geological carbon cycle and is a primary driver of the Sr-isotopic composition of seawater. Earth and Planetary Science Letters, 415, 38–46. https://doi.org/10.1016/j.epsl.2015.01.027
Coogan, Laurence A., & Gillis, K. M. (2013). Evidence that low-temperature oceanic hydrothermal systems play an important role in the silicate-carbonate weathering cycle and long-term climate regulation. Geochemistry, Geophysics, Geosystems, 14(6), 1771–1786. https://doi.org/10.1002/ggge.20113
Driese, S. G., Jirsa, M. A., Ren, M., Brantley, S. L., Sheldon, N. D., Parker, D., & Schmitz, M. (2011). Neoarchean paleoweathering of tonalite and metabasalt: Implications for reconstructions of 2.69Ga early terrestrial ecosystems and paleoatmospheric chemistry. Precambrian Research, 189(1), 1–17. https://doi.org/https://doi.org/10.1016/j.precamres.2011.04.003
Falkowski, P. G., Katz, M. E., Milligan, A. J., Fennel, K., Cramer, B. S., Aubry, M. P., et al. (2005). The Rise of Oxygen over the Past 205 Million Years and the Evolution of Large Placental Mammals. Science, 309(5744), 2202 LP – 2204. https://doi.org/10.1126/science.1116047
Farquar, Bao, & Thiemens. (2000). Atmospheric Influence of Earth’s Earliest Sulfur Cycle. Science, 289(5480), 756–758.
Friedrich, O., Norris, R. D., & Erbacher, J. (2012). Evolution of middle to Late Cretaceous oceans—A 55 m.y. record of Earth’s temperature and carbon cycle. Geology, 40(2), 107–110. https://doi.org/10.1130/G32701.1
Giesen, A., Herzler, J., & Roth, P. (2002). High temperature oxidation of iron atoms by CO2. Physical Chemistry Chemical Physics, 4(15), 3665–3668. https://doi.org/10.1039/B201822E
Goldblatt, Claire, Lenton, Matthews, Watson, & Zahnle. (2009a). Nitrogen-enhanced greenhouse warming on early Earth. Nature Geoscience, 2.
Goldblatt, C. (2016). Comment on “Long-term climate forcing by atmospheric oxygen concentrations.” Science, 353(6295), 132 LP – 132. https://doi.org/10.1126/science.aad6976
106
Goldblatt, C., Claire, M. W., Lenton, T. M., Matthews, A. J., Watson, A. J., & Zahnle, K. J. (2009b). Nitrogen-enhanced greenhouse warming on earlyEarth. Nature Geoscience, 2(12), 891–896. https://doi.org/10.1038/ngeo692
Gough, D. O. (1981). Solar Interior Structure and Luminosity Variations BT - Physics of Solar Variations. In V. Domingo (Ed.) (pp. 21–34). Dordrecht: Springer Netherlands.
Haqq-Misra, J. D., Domagal-Goldman, S. D., Kasting, P. J., & Kasting, J. F. (2008). A revised, hazy methane greenhouse for the Archean Earth. Astrobiology, 8(6), 1127–1137. https://doi.org/10.1089/ast.2007.0197
Harman, Schwieterman, Schottlekotte, & Kasting. (2015). Abiotic O2 levels on planets around F, G, K, and M stars: possible false positives for life? The Astrophysical Journal, 812(2).
Harman, C. E., Pavlov, A. A., Babikov, D., & Kasting, J. F. (2018). Chain formation as a mechanism for mass-independent fractionation of sulfur isotopes in the Archean atmosphere. Earth and Planetary Science Letters, 496, 238–247. https://doi.org/10.1016/j.epsl.2018.05.041
Hetherington, A. J., & Dolan, L. (2019). Rhynie chert fossils demonstrate the independent origin and gradual evolution of lycophyte roots. Current Opinion in Plant Biology, 47, 119–126. https://doi.org/https://doi.org/10.1016/j.pbi.2018.12.001
Holland. (1994). Early life on Earth. In Nobel Symposium, Vol. 4 (pp. 237–244). Columbia University Press.
Holland. (2006). The oxygenation of the atmosphere and oceans. Phil. Trans. of the Royal Society – Biological Sciences, 361, 903–915.
Holland, M. M., & Bitz, C. M. (2003). Polar amplification of climate change in coupled models. Climate Dynamics, 21(3–4), 221–232. https://doi.org/10.1007/s00382-003-0332-6
Intergovernmental Panel on Climate Change (Ed.). (2014). Anthropogenic and Natural Radiative Forcing. In Climate Change 2013 – The Physical Science Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (pp. 659–740). Cambridge: Cambridge University Press. https://doi.org/DOI: 10.1017/CBO9781107415324.018
Isson, T. T., & Planavsky, N. J. (2018). Reverse weathering as a long-term stabilizer of marine pH and planetary climate. Nature, 560(7719), 471–475. https://doi.org/10.1038/s41586-018-0408-4
Kah, L. C., & Riding, R. (2007). Mesoproterozoic carbon dioxide levels inferred from calcified cyanobacteria. Geology, 35(9), 799–802. https://doi.org/10.1130/G23680A.1
107
Kanzaki, Y., & Murakami, T. (2015). Estimates of atmospheric CO2 in the Neoarchean–Paleoproterozoic from paleosols. Geochimica et Cosmochimica Acta, 159, 190–219. https://doi.org/https://doi.org/10.1016/j.gca.2015.03.011
Kasting. (2014). Atmospheric composition of Hadean−early Archean Earth: The importance of CO. Geological Society of America Special Papers, 504, 19–28.
Kasting, J.F. (2010). Faint young sun redux. Nature, 464, 687–689.
Kasting, J F. (1993). Earth's early atmosphere. Science, 259(5097), 920 LP – 926. https://doi.org/10.1126/science.11536547
Kasting, James F. (1984). The evolution of the prebiotic atmosphere. Origins of Life, 14, 75–82.
Kasting, James F. (1987). Theoretical constraints on oxygen and carbon dioxide concentrations in the Precambrian atmosphere. Precambrian Research, 34(3), 205–229. https://doi.org/https://doi.org/10.1016/0301-9268(87)90001-5
Kharecha, P., Kasting, & Siefert. (2005). A coupled atmosphere–ecosystem model of the early Archean Earth. Geobiology, 3(2), 53–76. https://doi.org/10.1111/j.1472-4669.2005.00049.x
Kidston, R., & Lang, W. H. (1921). XXXIII.—On Old Red Sandstone Plants showing Structure, from the Rhynie Chert Bed, Aberdeenshire. Part V. The Thallophyta occurring in the Peat-Bed; the Succession of the Plants throughout a Vertical Section of the Bed, and the Conditions of Accumulation a. Transactions of the Royal Society of Edinburgh, 52(4), 855–902. https://doi.org/DOI: 10.1017/S0080456800016045
Kiehl, J. T., Hack, J. J., Bonan, G. B., Boville, B. A., Williamson, D. L., & Rasch, P. J. (1998). The National Center for Atmospheric Research Community Climate Model: CCM3*. Journal of Climate, 11(6), 1131–1149. https://doi.org/10.1175/1520-0442(1998)011<1131:TNCFAR>2.0.CO;2
Kirschvink, J. L., Gaidos, E. J., Bertani, L. E., Beukes, N. J., Gutzmer, J., Maepa, L. N., & Steinberger, R. E. (2000). Paleoproterozoic snowball Earth: Extreme climatic and geochemical global change and its biological consequences. Proceedings of the National Academy of Sciences, 97(4), 1400 LP – 1405. https://doi.org/10.1073/pnas.97.4.1400
Knutti, R., Rugenstein, M. A. A., & Hegerl, G. C. (2017). Beyond equilibrium climate sensitivity. Nature Geoscience, 10(10), 727–736. https://doi.org/10.1038/ngeo3017
Kopp, R. E., Kirschvink, J. L., Hilburn, I. A., & Nash, C. Z. (2005). The Paleoproterozoic snowball Earth: A climate disaster triggered by the evolution of oxygenic photosynthesis. Proceedings of the National Academy of Sciences of the United States of America, 102(32), 11131 LP – 11136. https://doi.org/10.1073/pnas.0504878102
108
Krissansen-Totton, J., & Catling, D. C. (2017). Constraining climate sensitivity and continental versus seafloor weathering using an inverse geological carbon cycle model. Nature Communications, 8(May), 1–15. https://doi.org/10.1038/ncomms15423
Kuhn, W. R., & Atreya, S. K. (1979). Ammonia photolysis and the greenhouse effect in the primordial atmosphere of the earth. Icarus, 37(1), 207–213. https://doi.org/https://doi.org/10.1016/0019-1035(79)90126-X
Kump, & Arthur. (1997). Global Chemical Erosion during the Cenozoic: Weatherability Balances the Budgets. In Tectonic Uplift and Climate Change (pp. 399–426). Springer.
Kump, Kasting, & Crane. (2010). The Earth System (3rd ed.). Pearson.
Kump, L. R., Brantley, S. L., & Arthur, M. A. (2000). Chemical Weathering, Atmospheric CO2, and Climate. Annual Review of Earth and Planetary Sciences, 28(1), 611–667. https://doi.org/10.1146/annurev.earth.28.1.611
Lambertz, M., Bertozzo, F., & Sander, P. M. (2018). Bone histological correlates for air sacs and their implications for understanding the origin of the dinosaurian respiratory system. Biology Letters, 14(1), 20170514. https://doi.org/10.1098/rsbl.2017.0514
Lehmer, O. R., Catling, D. C., Buick, R., Brownlee, D. E., & Newport, S. (2020). Atmospheric CO<sub>2</sub> levels from 2.7 billion years ago inferred from micrometeorite oxidation. Science Advances, 6(4), eaay4644. https://doi.org/10.1126/sciadv.aay4644
Lenton, T. M., Daines, S. J., & Mills, B. J. W. (2018). COPSE reloaded: An improved model of biogeochemical cycling over Phanerozoic time. Earth-Science Reviews, 178(August 2017), 1–28. https://doi.org/10.1016/j.earscirev.2017.12.004
Li, C.-W., Chen, J.-Y., & Hua, T.-E. (1998). Precambrian Sponges with Cellular Structures. Science, 279(5352), 879 LP – 882. https://doi.org/10.1126/science.279.5352.879
Li, K.-F., Pahlevan, K., Kirschvink, J. L., & Yung, Y. L. (2009). Atmospheric pressure as a natural climate regulator for a terrestrial planet with a biosphere. Proceedings of the National Academy of Sciences, 106(24), 9576 LP – 9579. https://doi.org/10.1073/pnas.0809436106
Love, S. G., & Brownlee, D. E. (1991). Heating and thermal transformation of micrometeoroids entering the Earth’s atmosphere. Icarus, 89(1), 26–43. https://doi.org/10.1016/0019-1035(91)90085-8
Lyons, S. L., Baczynski, A. A., Babila, T. L., Bralower, T. J., Hajek, E. A., Kump, L. R., et al. (2019). Palaeocene–Eocene Thermal Maximum prolonged by fossil carbon oxidation. Nature Geoscience, 12(1), 54–60. https://doi.org/10.1038/s41561-018-
109
0277-3
Macdonald, F. A., Swanson-Hysell, N. L., Park, Y., Lisiecki, L., & Jagoutz, O. (2019). Arc-continent collisions in the tropics set Earth’s climate state. Science, 364(6436), 181 LP – 184. https://doi.org/10.1126/science.aav5300
Maher, K., & Chamberlain, C. P. (2014). Hydrologic Regulation of Chemical Weathering and the Geologic Carbon Cycle. Science, 343(6178), 1502 LP – 1504. https://doi.org/10.1126/science.1250770
Manabe, S., & Wetherald, R. T. (1967). Thermal Equilibrium of the Atmosphere with a Given Distribution of Relative Humidity. Journal of the Atmospheric Sciences, 24(3), 241–259. https://doi.org/10.1175/1520-0469(1967)024<0241:TEOTAW>2.0.CO;2
Marty, B., Zimmermann, L., Pujol, M., Burgess, R., & Philippot, P. (2013). Nitrogen Isotopic Composition and Density of the Archean Atmosphere. Science, 342(6154), 101 LP – 104. https://doi.org/10.1126/science.1240971
Massie, S. T., & Hunten, D. M. (1981). Stratospheric eddy diffusion coefficients from tracer data. Journal of Geophysical Research: Oceans, 86(C10), 9859–9868. https://doi.org/10.1029/JC086iC10p09859
Mlawer, E. J., Taubman, S. J., Brown, P. D., Iacono, M. J., & Clough, S. A. (1997). Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. Journal of Geophysical Research: Atmospheres, 102(D14), 16663–16682. https://doi.org/10.1029/97JD00237
Mojzsis, S. J., Harrison, T. M., & Pidgeon, R. T. (2001). Oxygen-isotope evidence from ancient zircons for liquid water at the Earth’s surface 4,300 Myr ago. Nature, 409(6817), 178–181. https://doi.org/10.1038/35051557
Mora, C. I., Driese, S. G., & Seager, P. G. (1991). Carbon dioxide in the Paleozoic atmosphere: evidence from carbon- isotope compositions of pedogenic carbonate. Geology, 19(10), 1017–1020. https://doi.org/10.1130/0091-7613(1991)019<1017:CDITPA>2.3.CO;2
Müller, R. D., Sdrolias, M., Gaina, C., Steinberger, B., & Heine, C. (2008). Long-Term Sea-Level Fluctuations Driven by Ocean Basin Dynamics. Science, 319(5868), 1357 LP – 1362. https://doi.org/10.1126/science.1151540
Nutman, A. P., Bennett, V. C., Friend, C. R. L., Van Kranendonk, M. J., & Chivas, A. R. (2016). Rapid emergence of life shown by discovery of 3,700-million-year-old microbial structures. Nature, 537(7621), 535–538. https://doi.org/10.1038/nature19355
Ojakangas, Srinivasan, Hegde, Chandrakant, & Srikantia. (2014). The Talya Conglomerate: an Archean (∼2.7 Ga) Glaciomarine Formation, Western Dharwar Craton, Southern India. Current Science, 106(3).
110
Pavlov, A. A., Brown, L. L., & Kasting, J. F. (2001). UV shielding of NH3 and O2 by organic hazes in the Archean atmosphere. Journal of Geophysical Research, 106(106), 23267–23287. https://doi.org/10.1029/2000JE001448
Poulsen, C. J., & Zhou, J. (2013). Sensitivity of Arctic Climate Variability to Mean State: Insights from the Cretaceous. Journal of Climate, 26(18), 7003–7022. https://doi.org/10.1175/JCLI-D-12-00825.1
Poulsen, C. J., Tabor, C., & White, J. D. (2015). Long-term climate forcing by atmospheric oxygen concentrations. Science, 348(6240), 1238 LP – 1241. https://doi.org/10.1126/science.1260670
Quick, D. E., & Ruben, J. A. (2009). Cardio-pulmonary anatomy in theropod dinosaurs: Implications from extant archosaurs. Journal of Morphology, 270(10), 1232–1246. https://doi.org/10.1002/jmor.10752
Raymo, M. E., & Ruddiman, W. F. (1992). Tectonic forcing of late Cenozoic climate. Nature, 359(6391), 117–122. https://doi.org/10.1038/359117a0
Ribas, I., Guinan, E. F., Gudel, M., & Audard, M. (2005). Evolution of the Solar Activity over Time and Effects on Planetary Atmospheres. I. High‐Energy Irradiances (1–1700 A). The Astrophysical Journal, 622(1), 680–694. https://doi.org/10.1086/427977
Rimmer, P. B., Shorttle, O., & Rugheimer, S. (2019). Oxidised micrometeorites as evidence for low atmospheric pressure on the early Earth. Geochemical Perspectives Letters, 9, 38–42. https://doi.org/10.7185/geochemlet.1903
Rollason, R. J., & Plane, J. M. C. (2000). The reactions of FeO with O3, H2 , H2O, O2 and CO2. Physical Chemistry Chemical Physics, 2(10), 2335–2343. https://doi.org/10.1039/B000877J
Rothman, D. H. (2002). Atmospheric carbon dioxide levels for the last 500 million years. Proceedings of the National Academy of Sciences of the United States of America, 99(7), 4167–4171. https://doi.org/10.1073/pnas.022055499
Royer, D. L. (2013). Atmospheric CO2 and O2 During the Phanerozoic: Tools, Patterns, and Impacts. Treatise on Geochemistry: Second Edition (2nd ed., Vol. 6). Elsevier Ltd. https://doi.org/10.1016/B978-0-08-095975-7.01311-5
ROYER, D. L., & HREN, M. T. (2017). CARBON ISOTOPIC FRACTIONATION BETWEEN WHOLE LEAVES AND CUTICLE. PALAIOS, 32(4), 199–205. https://doi.org/10.2110/palo.2016.073
Royer, Dana L, Donnadieu, Y., Park, J., Kowalczyk, J., & Goddéris, Y. (2014). Error analysis of CO2 and O2 estimates from the long-term geochemical model GEOCARBSULF. American Journal of Science , 314(9), 1259–1283. Retrieved from http://www.ajsonline.org/content/314/9/1259.abstract
111
Sagan, C., & Mullen, G. (1972). Earth and Mars: Evolution of Atmospheres and Surface Temperatures. Science, 177(4043), 52 LP – 56. https://doi.org/10.1126/science.177.4043.52
Saltzman, M. R., & Edwards, C. T. (2017). Gradients in the carbon isotopic composition of Ordovician shallow water carbonates: A potential pitfall in estimates of ancient CO2 and O2. Earth and Planetary Science Letters, 464, 46–54. https://doi.org/https://doi.org/10.1016/j.epsl.2017.02.011
Sander, P. M., Christian, A., Clauss, M., Fechner, R., Gee, C. T., Griebeler, E.-M., et al. (2011). Biology of the sauropod dinosaurs: the evolution of gigantism. Biological Reviews of the Cambridge Philosophical Society, 86(1), 117–155. https://doi.org/10.1111/j.1469-185X.2010.00137.x
Scott, A. C., & Glasspool, I. J. (2006). The diversification of Paleozoic fire systems and fluctuations in atmospheric oxygen concentration. Proceedings of the National Academy of Sciences of the United States of America, 103(29), 10861–10865. https://doi.org/10.1073/pnas.0604090103
Segura, A., Krelove, K., Kasting, J. F., Sommerlatt, D., Meadows, V., Crisp, D., et al. (2003a). Ozone concentrations and ultraviolet fluxes on earth-like planets around other stars. Astrobiology, 3(4), 689–708. https://doi.org/10.1089/153110703322736024
Segura, A., Krelove, K., Kasting, J. F., Sommerlatt, D., Meadows, V., Crisp, D., et al. (2003b). Ozone Concentrations and Ultraviolet Fluxes on Earth-Like Planets Around Other Stars. Astrobiology, 3(4), 689–708. https://doi.org/10.1089/153110703322736024
Self, D. E., & Plane, J. M. C. (2003). A kinetic study of the reactions of iron oxides and hydroxides relevant to the chemistry of iron in the upper mesosphere. Physical Chemistry Chemical Physics, 5(7), 1407–1418. https://doi.org/10.1039/B211900E
Smirnov, V. N. (2008). Rate constant of the gas-phase reaction between Fe atoms and CO2. Kinetics and Catalysis, 49(5), 607–609. https://doi.org/10.1134/S0023158408050017
Smirnov, V. N. (2012). Thermochemical parameters and rate constants of the reactions Fe + O2 + M ↔ FeO2 + M and FeO + O2 ↔ FeO2 + O. Kinetics and Catalysis, 53(5), 543–553. https://doi.org/10.1134/S0023158412050138
Som, S. M., Buick, R., Hagadorn, J. W., Blake, T. S., Perreault, J. M., Harnmeijer, J. P., & Catling, D. C. (2016). Earth’s air pressure 2.7 billion years ago constrained to less than half of modern levels. Nature Geoscience, 9(6), 448–451. https://doi.org/10.1038/ngeo2713
Stanton, C. L., Reinhard, C. T., Kasting, J. F., Ostrom, N. E., Haslun, J. A., Lyons, T. W., & Glass, J. B. (2018). Nitrous oxide from chemodenitrification: A possible missing link in the Proterozoic greenhouse and the evolution of aerobic respiration.
112
Geobiology, 16(6), 597–609. https://doi.org/10.1111/gbi.12311
Stolper, D. A., Bender, M. L., Dreyfus, G. B., Yan, Y., & Higgins, J. A. (2016). A Pleistocene ice core record of atmospheric O<sub>2</sub> concentrations. Science, 353(6306), 1427 LP – 1430. https://doi.org/10.1126/science.aaf5445
Tappert, R., McKellar, R. C., Wolfe, A. P., Tappert, M. C., Ortega-Blanco, J., & Muehlenbachs, K. (2013). Stable carbon isotopes of C3 plant resins and ambers record changes in atmospheric oxygen since the Triassic. Geochimica et Cosmochimica Acta, 121, 240–262. https://doi.org/https://doi.org/10.1016/j.gca.2013.07.011
Tomkins, A. G., Bowlt, L., Genge, M., Wilson, S. A., Brand, H. E. A., & Wykes, J. L. (2016). Ancient micrometeorites suggestive of an oxygen-rich Archaean upper atmosphere. Nature, 533, 235–238. https://doi.org/10.1038/nature17678
Toon, O. B., McKay, C. P., Ackerman, T. P., & Santhanam, K. (1989). Rapid calculation of radiative heating rates and photodissociation rates in inhomogeneous multiple scattering atmospheres. Journal of Geophysical Research: Atmospheres, 94(D13), 16287–16301. https://doi.org/10.1029/JD094iD13p16287
Velbel, M. A. (1993). Temperature dependence of silicate weathering in nature: How strong a negative feedback on long-term accumulation of atmospheric CO2 and global greenhouse warming? Geology, 21(12), 1059–1062. https://doi.org/10.1130/0091-7613(1993)021<1059:TDOSWI>2.3.CO;2
Volk. (1987). Feedbacks between weathering and atmospheric CO2 over the last 100 million years. American Journal of Science, 287(8), 763–779.
Wade, D. C., Abraham, N. L., Farnsworth, A., Valdes, P. J., Bragg, F., & Archibald, A. T. (2019). Simulating the climate response to atmospheric oxygen variability in the Phanerozoic: a focus on the Holocene, Cretaceous and Permian. Climate of the Past, 15(4), 1463–1483. https://doi.org/http://dx.doi.org/10.5194/cp-15-1463-2019
Walker. (1993). Biogeochemical Cycles of Carbon on a Hierarchy of Time Scales. In Biogeochemistry of Global Change (pp. 3–28).
Walker, J. C. G., Hays, P. B., & Kasting, J. F. (1981). A negative feedback mechanism for the long-term stabilization of Earth’s surface temperature. Journal of Geophysical Research: Oceans, 86(C10), 9776–9782. https://doi.org/10.1029/JC086iC10p09776
West, A. J., Galy, A., & Bickle, M. (2005). Tectonic and climatic controls on silicate weathering. Earth and Planetary Science Letters, 235(1), 211–228. https://doi.org/https://doi.org/10.1016/j.epsl.2005.03.020
Wilde, S. A., Valley, J. W., Peck, W. H., & Graham, C. M. (2001). Evidence from detrital zircons for the existence of continental crust and oceans on the Earth 4.4 Gyr ago. Nature, 409(6817), 175–178. https://doi.org/10.1038/35051550
113
Wordsworth, R., & Pierrehumbert, R. (2013). Hydrogen-Nitrogen Greenhouse Warming in Earth's Early Atmosphere. Science, 339(6115), 64 LP – 67. https://doi.org/10.1126/science.1225759
Young, G. M., Brunn, V. von, Gold, D. J. C., & Minter, W. E. L. (1998). Earth’s Oldest Reported Glaciation: Physical and Chemical Evidence from the Archean Mozaan Group (∼2.9 Ga) of South Africa. The Journal of Geology, 106(5), 523–538. https://doi.org/10.1086/516039
Zahnle, K., Claire, M., & Catling, D. (2006). The loss of mass-independent fractionation in sulfur due to a Palaeoproterozoic collapse of atmospheric methane. Geobiology, 4(4), 271–283. https://doi.org/10.1111/j.1472-4669.2006.00085.x
Zahnle, K. J. (1986). Photochemistry of methane and the formation of hydrocyanic acid (HCN) in the Earth’s early atmosphere. Journal of Geophysical Research: Atmospheres, 91(D2), 2819–2834. https://doi.org/10.1029/JD091iD02p02819
Zahnle, Kevin, & Buick, R. (2016). Ancient air caught by shooting stars. Nature, 533(7602), 184–186. https://doi.org/10.1038/533184a
Vita Rebecca C. Payne
Education Dual-title PhD, Geosciences and Astrobiology Anticipated graduation: Dec 2020
Penn State University, University Park, PA Bachelor of Sciences, Geological Sciences May 2015
Brown University, Providence, RI Publications Payne, Brownlee, & Kasting (2020): “Oxidized micrometeorites suggest either high
pCO2 or low pN2 during the Neoarchean”. PNAS. Hayworth, Kopparapu, Haqq-Misra, Batalha, Payne, Foley, Ikwut-Ukwa, & Kasting
(2020): “Warming early Mars with climate cycling: the effect of CO2-H2 collision-induced absorption”. Icarus.
Payne, Britt, Chen, Kasting, & Catling (2016): “The response of Phanerozoic surface temperature to variations in atmospheric oxygen concentration”. JGR: Atmospheres, 121.
Research Experience Doctoral Research, Penn State University 2015–2020
Research Advisor: James F. Kasting. Penn State University Dissertation: Select problems in the evolution of Earth’s atmosphere and climate
Undergraduate Research, Brown University 2014–2015 Research Advisor: Jung-Eun Lee, Brown University Thesis: Precipitation Seasonality in Extreme Climates Using an Idealized GCM
Teaching Experience Graduate Writing Tutor, Penn State University Jan 2017 – present Teaching assistant at Penn State University
Astrobiology Fall 2019 Geology of the National Parks (online course) Spring 2017, Spring 2019 Introduction to the Earth System Fall 2015, Fall 2018 Global Water Cycle (Brown University) Spring 2015
Conferences EON-ELSI Astrobiology Winter School—Earth-Life Sciences Institute, Tokyo Jan 2018 NASA Astrobiology Summer School—NASA and Centro de Astrobiologia June 2017 AGU Fall Meeting, talks and posters 2014, 2016, 2018, 2019 Awards and Honors 2nd place, Pre-Comps Talks—51st Geosciences Graduate Colloquium 2018 Krynine Scholarship 2018, 2019 Institutional Roles Geosciences Dept Graduate Colloquium budget manager; co-chair 2017–2020; 2020